pee Bey iL hk ean 
 
 "4" 
 Veg i] Wich ub es 
 4 tate f 
 ii KG SUA by ws Tiel, ay ee 
 
 ie Star ah 
 pin hoa i ii sae rt 
 ie i eet eu iy 
 
 a 
 i ae ted 
 Hah 4 aa iit bi 
 aie! Ms at 
 4 a aed Gite ia Hy 
 1 ‘ 
 
 Paste WuiPe Gy v 
 foisth ied 
 8 y iy he: ae al 
 
 ee thegs 
 
 wh 
 F i dain vi 
 
 sy on 7m 
 Pe feats + 
 a4 east 
 tH a i 8 
 
 U et 7 al a 4 
 + peutery Ca «dj 
 ior Aa SR ye be beh ull 
 dah oh it ' yah 4 godt a 
 Asi rie ett 
 
 aye ae i Pt Pate 
 pera ne rl its ‘aio 
 
 i j 
 hades 
 Ast dele ee 
 
 ings 
 
 patie i je \ i ae Dit 
 ane iu He ig  h hate elite fk) 
 Ase! is Haare 
 
 on % ie i of abt 
 
 Tie, icicle i ae 
 a . eae 
 
 at ae 
 
 APR 14 
 
 hi Hh) pO 
 Mat ‘ be fash: 
 i iH, 
 
 ied 
 
 faded 
 
 fan ad HAY it ish 
 
 Hhipa 
 
 Mi it iy ae 
 ae P iit ihe 
 
 ati be Pate 
 ae 
 aA Bie 
 Lay 
 
 ie ce (J ae i iii 
 
 Ae sila 
 
 ¥ } 7 cay " fl tah ind 
 ay" WHT HSEA| ) jigedoteltbe te IAF se SH LES ran WL Hie Apple a i a i hs Hs bbe nd i} Aly 
 ii ti: i ol a, t ih AE, ft 1 thats th i i ‘al ‘ Brak j 4 he ne ip wie ts i ee Hath aia! 
 bho ty 4 Hit AV ivi 4 eit 4 esl Qo tete ohn a i) 
 
 ytiea} ae Svea Hi HH Wet hdr jab Guede! Nani ne Deore ve hh 4 
 REARING ee ia oh a ie es iE ee aan io 
 ORLA it Gaede neadeis in ay wih i Ws belt il abil ihegets 4 {li taht re 4 feta 4 vic whadaileh Hi 
 MORO Ma WUE TR aL A ‘ iia ii de aah He Fy 
 ARRAY 4 AGA nu iy Ms) ft ws, Wit ten BB iy a ty 4 HH 
 
 fh f i" (ing 4 4 I: 8 Ths (hale i Pclba hy ru } : ‘46 
 4) ena i} bib i it A oe ae i ie ce il H ial 4 ihe | 
 i a i pe ri ue mi iets } Gctrecrarg bebo 
 ibeay ie ih at } iia! re pai i i hy 
 
 at fh. 19; ate 4 a I 
 
 a fo oe a ae 
 ] i) int (i HN Nie i ia ibaa AO 
 ‘ anh ine) oma tu Wy hy : sate Ny ynak i a a ae Vit iigiael Dt ti 
 en AE id ae iy i eae PAR Wi sO ae HOU rh A at ik vi ise i a ae SHEEN sate sity nit ‘ant 
 Oe ane EN aM Mi) OU es) 
 a that HOR Peer MESURE 
 Hi Henan ER 
 ie a 7 an ea ea est 
 i ae sue Ati hohe Fitna aah eT HO a HL | 
 oF s Ri eae ae 
 it i % i vio ue ite i ie oe ney a ee ue ratty 
 ii Mat tin ith i: iil He ie tae f js ti aN iit Bry 
 \ wath BS yaad MR ory 
 Aaron ie ee a Pans ea 
 
 q i } th i) noi My vk Wy 
 uy hy i a a at na Ht V 14 eat ie it Ht wn i a Ls iii if ig ) i ini 
 i q i Pac SL EAL bi FAO sitpo abl ncuconm danse viz 
 ry Reyer tae i HA sha iene Pak ri ade ce | ua eet ; ifiie maint “4 iit ant 2 uae 
 } tat btonsate abate ake 8 sted ih in ‘3 int batt rate iret a 
 ' aby it i ide coe i is ei “ahi i} Mien an Ht iy i BUH wi wi TRH ‘ i 
 PNW ONT Le ay i My f ain i tise ie a is) Watt a “Get 
 te ee ae Bs da iy “i HF iy! ti aiin ag) ih iad veh eat Hie) ie Hf 
 Tea eR iy bots} eth hb, ini Tea it ait an iia alsheia i 
 a en See ala ae 
 LL arate e ys be ait i Pe wi ae ri agit a’ a Aue He i fies i a iF fie PH a Alt : 
 i eas oy ey J Dy nd Hn va i i i : a ae 
 
 2 . i zn ct Mh i 4 ai 
 
 i 3 
 
 ris 
 
 i 
 i ih ds Hy a sf si ite 
 a. 
 iit a ‘t ie Hh 
 
 vi Nia} | 
 ; uh 4 
 
 Ait Wvid 
 SON os) Ran 
 
 : ns 
 va ? Mitel 
 
 ' H ie Ms ij srk ta AH i h 
 
 Hehe aie ate ‘i rt apna 
 
 } j diet) ay ENG ' i ity i Int ebay A ts shia a x taht ae Has i j y\ ite eee 
 "i ‘ cnn rear ts babe tt, So aitate AAW dit ie vii isn ing iN Haat i A Aeon i 
 in ai shlay i a : ia 4 ei a ' HL Ya it ia My ata trai ny ; Hi a Tbe oe oh me uch 3 ata 
 Nei wi aHAN Tatty i ita ad) AW MRR TELE EYRHE Wii pias iso ii alata hai 
 oH ¥ 4 i ii at Ai sath ith i ie Sea Ms My) he i iM cane 4 
 
 pai 
 on 
 
 wt ants oh RIEL 
 i inns i ce ei M4 Al Haan oc ae 
 
 hae ti : 
 ish ee i638 Hy Psi - 
 ih ee pa ae ie 
 
 ‘ AMR i int a i ‘ . nuke 
 
 a 
 
 . i aes i 
 
 t+ 
 sf q i weit 
 
 ti 
 ck anit Be rt i ie 
 Hens PE ae i lit ‘it a 
 i ee hes i taht nba Git : a 
 Ronnies 
 
 Rut mht LA Rap Fea 
 a bake it ura ih ih #) iN 
 H ae fedsnel i 
 Hejchites Pet at ae ln ie eh 
 nh i us — the ne — sn a 
 
 ipa wile . eta! 
 ine My ss ae ahi eae ‘i 
 ‘ H 
 
 MateRead 
 ely 
 ie . 
 
 aN 
 Why 4, Mitek 0 
 wey 
 THAT) 159 0 
 DG eae bed 
 
 is rie ra 
 
 et " ney 
 
 ata heh 
 nia MT tay 
 Died aans at 
 
 te , Ls 
 i « “ iit 
 cia iF BA 
 
 Mets net i vi fe 3 
 
 Ha 1 ut 
 
 . eb 
 Mili? i 
 Paigisihst * 
 
 Math my ! seit ra oy ut i 
 i . aa ae 
 it 
 
 inne heb hr 
 
 a sats ‘? Br fettede y Nery Meet 
 fee oA he ted 
 Maa} Ns dae A itu Neieh 
 
 he Vere de tek tee 
 
 TAHA I 
 Hibs Hs in 
 
 itp lat 
 i Ht nt 
 VO d 
 
 Dota i ol 
 Be a anes wat 
 
 Bi 
 My my ie ae Has at 
 a Hf iy a a. iH lye ANY 
 
 i Fhe ot 
 ‘ it, Hele bs oe Mi th 
 Wns Rk aH tite siete it" bala 
 ar Wii 4 nacht Hit i gigas rah igee 
 ? rt te af 
 
 ; i ih 
 tit Hila a Hsiek 
 ie a mates anal) | 
 
 ee a te 
 eae ot A At doko de eet 
 UA UR eS 
 ety Naty ded JOA ALS. 
 Cite on We ne Pc} APN ey 
 eA Ah Roh ite Reg Yee tae Dale 
 ar A benthic 8 59: 
 
 9) ym 9 2 ee eo 
 
 ait Aidt 
 fi nary 
 ; 4 
 wh 1 
 ae 
 sini ke K 
 rth) 
 4 Ben ra 3 ae ; i 
 (ehete tad tbe 
 tet lati: te ah 4 he ies it n 
 
 na Vekey 
 vitidetay ails 
 A 
 
 ae si ith oF i ie He ait | : 
 ah oe 
 
 ae i ive on at ae 
 
 tity" LT Ba 
 
 tau iit 
 at 
 
 , a) 
 oy c an ty oy ne q i 
 Deleas : soba Celt abe ASI, . 4 
 oh os tte r.4 matey . \" ihe nh ‘ it i) y nahn! ists v Na! Pues ft si uti ies it) H gta 
 Vite ny Is) FWeiks 1s *{ Pe Heth f iy re. sa th ee at y ) beyetee 
 wie biihek, ( ah 2A ae haha diWaone tet stots 
 aw ohearas te 1 { aan ately me ‘ oN i carne Oca mye 
 shee bei alle china. staat fh Set AYA ii Pua Nisgqesen hited t 
 4 b9. e a es " ran pe | Ait ait if aera “af ( aa! rt Wieite, is ; CrheDpePutiny 
 SMH. si Lt 1: a Laas 1 ett NA a sant A nigite " anes \ 
 
 Ni sant Meda acaati hs 
 stay 4 tena 
 
 4 
 hy ahad 
 es tt ORR tit oe r 
 Wire tA 
 
 Bae 
 77 Ai asad at 
 
 49, yeh fhe fa ae 
 ateh a 
 
MATHEMATICS 
 LIBRARY 
 
 as 
 
 %e 
 
Return this book on or before the 
 Latest Date stamped below. 
 
 University of Illinois Library 
 
 JUN 4 1964 
 JUL 1 1% 
 
 L161—H41 
 
Digitized by the Internet Archive 
 in 2022 with funding from 
 University of Illinois Urbana-Champaign 
 
 https://archive.org/details/treatiseonanalytOOhyme_0O 
 
ANALYTICAL GEOMETRY 
 
 , 
 
 THREE DIMENSIONS. 
 
 B 
 
A TREATISE 
 
 ON 
 
 ANALYTICAL GEOMETRY 
 THREE DIMENSIONS, 
 
 CONTAINING THE 
 
 THEORY OF CURVE SURFACKS, 
 
 rpAND,OF. ,, 
 
 CURVES OF DOUBLE CURVATURE. 
 
 BY J. HYMERS, D.D. 
 
 FELLOW AND TUTOR OF ST JOHN’S COLLEGE, CAMBRIDGE. 
 
 THIRD EDITION, ALTERED AND REVISED. 
 
 CAMBRIDGE: 
 PRINTED AT THE UNIVERSITY PRESS, 
 FOR 
 J.& J.J. DEIGHTON, AND MACMILLAN & CO., CAMBRIDGE. 
 WHITTAKER & CO., LONDON. 
 
 M.DCCC.XLVIII. 
 
eats (Hild Ad 
 (yaaa TA AI Ds ee a 
 
 ~~ 2 
 i SORA Ma) hd % Vis yy a ee ie 
 : “ 
 . Mia red me Gi "1, ga 7 
 
 eign? Seo 
 
Ble 
 
 Li PRACt fT 
 
 te 
 
 Oa 
 
 ; 4 tia, o* 
 , J om i 
 } sacl ie oe 
 j 3 | / 
 7 g } \ 
 V 
 
 } wr 7 
 
 'y ; 12 An 
 i$ fs e ¢ 
 
 t _Y 
 
 CONTENTS. 
 
 a 
 i 
 t 
 
 ‘of 
 é 
 
 d 
 
 SECTION LI. 
 
 ON THE PLANE AND STRAIGHT LINE. 
 
 ART, PAGE 
 1—16. Equations to a plane and straight line............... 1 
 -17—41.. Problems on the plane and straight line............... 18 
 
 SECTION ILI. 
 ON SURFACES OF THE SECOND ORDER. 
 
 42—-52. Equations to a sphere, and to a cylinder, cone, and 
 BH TIRCe COTE YUM ence ae rr iste ate dng soa e eaten pin tee cats 32 
 
 53—65. Equations and forms of surfaces that have a center.... 39 
 
 66—72. Equations and forms of surfaces that have not a 
 EEC ee eee OCT AG. coe oiet Sth een EN ha Ot ay CRE ol 55 cA 50 
 
 SECTION III. 
 
 ON THE PROJECTIONS OF LINES AND PLANE SURFACES, AND ON 
 THE TRANSFORMATION OF CO-ORDINATES. 
 
 79—82. Length of the projection of a line, and area of the 
 Rrorccholm Ole Ay DIANE sSUtt acts eee cats eo eit ce vb ps as SRR 
 
 Beene) PL OU CUCL CO-OLUINALCS niwerer: svar copes Scr cle ss a's se 64 
 88—98. Transformation of co-ordinates. Polar co-ordinates... 68 
 
 99—106. Plane sections of surfaces............. S25 snk Some: a2..0070 
 
vi CONTENTS. 
 
 SECTION IV. 
 
 ON TANGENT PLANES AND NORMALS TO CURVE SURFACES, AND 
 THEIR VOLUMES AND AREAS. 
 
 ART. PAGE 
 
 107—111. Equations to the tangent plane and normal......... 86 
 
 112—120. Differential coefficients of the volume and area of a 
 curve surface in rectangular and polar co-ordinates ....... 90 
 
 SECTION V. 
 
 ON TANGENT LINES AND NORMAL AND OSCULATING PLANES TO 
 CURVES, AND THEIR LENGTHS. 
 
 121—130. Method of assigning the form of a curve in space 
 by equations. Tangent line, and normal and osculating 
 
 planes..... eR eer ea roe ee re a 96 
 
 131—132. Differential coefficient of the arc of a curve in rect- 
 angular or polar co-ordimates...c24 ae. 2 eens oe 104 
 
 SECTION VI. 
 
 ON THE DISCUSSION OF THE GENERAL EQUATION OF THE SECOND 
 
 ORDER. 
 
 133—138. Center, and diametral planes of a surface......... 106 
 139—144. Reduction of the general equation of the second 
 
 Oe Ee hac Fe ers Pre cat A RE Nya ony eee eee 112 
 145—148. Mode of determining the species and form of the 
 
 surface represented by a proposed equation...... Ait AcioN 118 
 149—150. Case of surfaces of revolution.............c.cc00- 2 ie 
 151—155. Properties of conjugate diametral planes............ 1 
 
 156—157. Position and magnitude of paraboloids, when repre- 
 sented by the general equation of the second order....... 133 
 
CONTENTS. Vil 
 
 SECTION VII. 
 
 ON CYLINDRICAL, CONICAL, AND CONOIDAL SURFACES, AND ON 
 
 SURFACES OF REVOLUTION. 
 ART, PAGE 
 
 158—159. Number of constants in the equations to surfaces. 
 Mode of representing the intersections of surfaces...... 137 
 
 160—172. Finite and differential equations of cylindrical, coni- 
 cal, and conoidal surfaces ; and surfaces of revolution...... 141 
 
 SECTION VIII. 
 
 ON SURFACES HAVING MORE THAN ONE ARBITRARY CURVILINEAR 
 DIRECTRIX, AND ON DEVELOPABLE SURFACES, AND ENVELOPES. 
 
 173—177. Twisted surfaces generated by a straight line that 
 constantly passes through two curves and remains parallel 
 to a fixed plane, or which constantly passes through three 
 
 BIVVCUACUL VES ty giaoate fect ne clei stain tare) Ge ae sib aries oe it'6 6 148 
 Ps =10l-sDevelopable Surtaces eras ta ess ake te ecko sine ass « 153 
 TOS eaol OG MORVElOPes at te hears -reres pety a icler ahr ssa «amie fee 167 
 
 SECTION IX. 
 ON THE CURVATURES OF CURVES IN SPACE. 
 
 197—201. Intersection of consecutive normal planes of a curve. 172 
 
 202—206. Radius of spherical curvature ; radius of absolute 
 CLE CULES oon ee oa Scone te ee Pe ot en iets eee Ti elchs a bhe alate te eNetete oles 174 
 
 907—210. Evolutes. of a curve in*space.....:.....2..se00s.0: 178 
 
 SECTION X. 
 ON THE CURVATURE OF SURFACES. 
 
 211—214. Conditions for a contact of any order between two 
 THEIACESUR. ties she < BPO be DOC OOOO Pare Pees ecatet a a's . 181 
 
Vill CONTENTS. 
 
 ART. PAGE 
 215—218. Radii of curvature of oblique and normal sections... 184 
 219—226. Normal Sections of greatest and least curvature... 189 
 227—229. Intersection of consecutive normals of a surface... 197 
 
 230—237. Lines of curvature, Principal radii of curvature, and 
 RMpilict of surlaces 4s: ine acre ksl. daahs Mein 5 2 AG 200 
 
 PPPODIEINS cts ciate hk 8a he lin pa ce ete hie ales 3 ae ee 209 
 
 Students reading this subject for the first time may confine their 
 attention to the following Articles. 
 
 1—41, 42—52, 57—71, 79—84, 107, 109—115, 121—126, 131. 
 
 ERRATA. 
 PAGE LINE ERROR. CORRECTION. 
 107% 3 YR LL 
 21 22 before this assertion supply the converse of 
 109 15 L,Y, z h, k,l 
 118 ] vy HS 
 121 9 +75+38 +255 -9 
 so that the surface is an ellipsoid. 
 133 12 2A’ x 2 A’, x 
 136 5 (> 1) ey kh) 
 7 a 
 dr dr 
 19] last line Peps h tan? 6 i7;:? 
 205 24 +4uv +4p?g?uv 
 
 which causes the two succeeding equations to be incorrect; the 
 equation at the top of p. 206 should be 
 2p? g° (1+ Raa +92)—p292 
 lip)e-(149%=" a, (L+p?)(1 +9°)-p°¢" 
 p”) ( +9° 1+ p? u f+ (l+p?)? 4u*p?9?=0, 
 231 li (rrr?) (rr'r’)2 
 251 1 : a gn 
 
ANALYTICAL GEOMETRY 
 
 OF 
 
 THREE DIMENSIONS. 
 
 SE GDTON I. 
 
 ON THE PLANE, AND STRAIGHT LINE. 
 
 1. Iw order to determine the position of a point in space, 
 some fixed point is taken for the origin of co-ordinates, and 
 through it are drawn three fixed planes, called the co-ordinate 
 planes, at right angles to one another, and intersecting one 
 another two and two in straight lines, which are also at right 
 angles to one another, and are called the awes of the co- 
 ordinates. Then, if the perpendicular distances of a point 
 from each of the co-ordinate planes be given, its position will 
 be completely determined. 
 
 Let O (fig. 1) be the origin of the co-ordinates, and y Oz, 
 Ov, wOy the co-ordinate planes; M any point, and ME, 
 MF, MN the perpendiculars let fall from it upon the co- 
 ordinate planes; these perpendiculars are called the rectan- 
 gular co-ordinates of M, and, as their values change for the 
 different points of space, are denoted by the variables a, y, z. 
 The point WM will be determined in position, if we know the 
 values of its three co-ordinates; that is, if we know that for 
 that point 
 
 Oe 0.8 SC, 
 
 For, if along Ov we measure OA =a, and through A draw 
 an indefinite plane parallel to yOz, this plane will contain all 
 points whose distance from yOx is a, or for which w=a, 
 
 1 
 
2 
 
 and, therefore, the point in question. Similarly, if we mea- 
 sure along Oy, Ox, the distances OB=b, OC =c, and 
 through B, C, draw indefinite planes respectively parallel 
 to Ox, wOy, each of these must contain the point in 
 question; therefore the three planes will, by their intersection 
 in M, determine one single position for the point whose 
 co-ordinates are w =a, y=b, x =c; which position coincides, 
 as we see, with the angular point, opposite to the origin, of 
 the rectangular parallelopiped constructed upon the edges 
 OA, OB, OC, equal to the three given co-ordinates. 
 
 2. The feet of the perpendiculars L, F, N, are called 
 the projections of the point M upon the co-ordinate planes. 
 Of these if any two be given, the third can be found; 
 thus if E, & be given, draw EB, FA, parallel to Oz; 
 then BN, AN, drawn respectively parallel to Ox, Oy, will 
 by their concourse determine JN. 
 
 Instead of the perpendiculars ME, MF, MN, the three 
 lines OA, AN, NM, which are respectively equal to them, 
 are commonly called the co-ordinates of M, and denoted by 
 @, Y, 3; also the axes of the co-ordinates Ox, Oy, Oz, 
 are often called the axis of w, the axis of y, and the axis 
 of x; and the co-ordinate planes vOy, yOx, xOw, the 
 planes of wy, yx, xa, respectively. 
 
 3. The determination of the point M will not however 
 be complete, unless we take into account the signs of the 
 quantities a, 6, c, in the equations 
 
 U =10e Yi Ds AAC 
 
 in order to measure these distances, when they are positive, 
 along the positive parts Ow, Oy, Ox, of the co-ordinate axes ; 
 or along the negative parts Ow’, Oy’, Oz’, of the axes pro- 
 duced in the contrary direction, when they are negative; as 
 is explained in Plane Geometry. For since the co-ordinate 
 planes, which must be supposed to be prolonged indefinitely, 
 form among themselves eight solid angles, there are eight 
 positions in which M might be situated at absolute distances 
 
3 
 
 a, b, c from the co-ordinate planes; and it is only by attending 
 to the signs that we shall be enabled to select the true one. 
 Without going through the combination of signs which belongs 
 to each of the solid angles, it may be sufficient to observe 
 that, if a point be situated in the compartment wys, all its 
 co-ordinates will be positive; if in the compartment a’y’x’, all 
 negative; and that for points in the compartments vy'z, a’y2’, 
 we must have, respectively, 
 
 t= dy, Y=—b, S=C¢; #=—a, Y=b, sie, 
 
 4, Sometimes it is requisite to take the co-ordinate 
 planes not at right angles, but inclined at given angles, to 
 one another, in which case the system of co-ordinates is 
 called oblique. 
 
 Thus (fig. 2) if wOy, yOx, sOw be three planes drawn 
 through the point O, and intersecting one another two and 
 two at given angles, in the lines Ov, Oy, Ox; and if from 
 any point M, MN be drawn parallel to Ox meeting the plane 
 vw Oy in N, and NA be drawn in that plane parallel to Oy ; 
 OA, AN, NM, are the co-ordinates of the point M referred 
 to the oblique axes Ox, Oy, Ox; and WM is called the pro- 
 jection of M made parallel to the axis Ox. 
 
 In the present chapter we shall confine our attention to 
 the case of rectangular co-ordinates; noticing however, as 
 they arise, several relations of lines and planes to one another, 
 the analytical expressions of which, and mode of investigation, 
 are the same, whether the axes of the co-ordinates be rect- 
 angular or oblique. 
 
 5. To find the distance of a point from the origin in 
 terms of its co-ordinates. 
 
 Let IW (fig. 1) be the point, 
 OA=2, AN=y’, NM =» its given co-ordinates ; 
 join OM, ON, and let OM = d. 
 Then from the triangle ONM, right-angled at N, 
 OM? = ON’? + NM’; 
 
 ra 
 
4 
 
 but the triangle OAN, right-angled at A, gives 
 ON* = OA’ + AN’; 
 “. OM? = OA? + AN’? + NM’; 
 
 Cor. Let a, 8, y be the angles which OM makes with 
 the axes of a, y, x; then, joining AM, since angle OAM is 
 a right angle, 
 
 OA x 
 
 cosa = —— = —; 
 
 OM da’ 
 .. # =dcosa; similarly, y’ =dcosB, zx’ =d cosy. 
 
 n r2 12 
 Also (cos a)’ + (cos (3)? + (cos vy)’ = a = 1. 
 This is the condition which the three angles, made by 
 a line through the origin with the co-ordinate axes, must 
 satisfy ; they cannot, therefore, be assumed at pleasure, but 
 two being given, a, §, for instance, the third is determined 
 by the equation 
 
 cosy = + 4/1 — (cosa)* — (cos 3)”, 
 
 and will therefore have two values, y and t—-y; the value 
 a — yy corresponding to a line OM’ in the plane OM, which 
 is inclined at an angle y to Oz’, and visibly makes angles 
 equal to a, B with Ow, Oy, the same as OM does. 
 
 6. To find the distance between two points in terms of 
 their co-ordinates. 
 
 Let M’ (fig. 3) be a point whose co-ordinates are x’, y’, 2’, 
 and NM any other point whose co-ordinates are wv, y, #3; join 
 . MM’ =d, and upon MM’, as diagonal, describe a rectangular 
 parallelopiped having its edges parallel to the co-ordinate axes. 
 
 Then from the triangle M’MK, right-angled at K, 
 MM” = M'K? + KM? = M’K? + (2 - 2’). 
 
5 
 
 But the triangle NW’LN, right-angled at L, gives 
 M'K? = N'N? = (#- af + (y-y')’; 
 MM? = (a#-2')? + (y-y) + (@- 2’), 
 or d=/ (wv — a’)? + (y—y)? + (#—8') 
 Both in this formula, and in that of Art. 5, we take the 
 
 radical with a positive sign; as the question only relates to 
 the absolute distance of the points. 
 
 Cor. Let a, 8, y be the angles which MM’ makes with 
 the edges of the parallelopiped, and which are manifestly the 
 same as those made by a line OP, drawn through the origin 
 parallel to MM’, with the co-ordinate axes. Then from the 
 
 triangle MM’H, right-angled at H, we have 
 MH «v-« 
 Mis adi i 
 
 cosa = 
 
 similarly, cos 3 -2? (Sttrigh Z—-z 
 
 7. To find the equation to a plane. 
 
 We may consider a plane as a surface generated by a 
 straight line which moves so as always to remain parallel 
 to itself, and to intersect a straight line given in position. 
 
 Let AC (fig. 5) in the plane of xa be the line to which 
 the generating line PQ is always parallel, and BC in the plane 
 of yz, intersecting the former in C, the line along which it 
 moves. Let s = 4v+ec be the equation to the line AC 
 referred to the axes Ov, Oz, in the plane in which it lies; 
 similarly, let s = By +c be the equation to BC referred to 
 the axes Oy, Ox, where in both cases c= OC. From P any 
 point in the generating line, and from Q where it meets BC, 
 draw PN, QM, parallel to the axis of z, and join MN. Then 
 the plane QN is parallel to zw, and therefore MN is parallel 
 to Ow (Euc. xt, 15, 16); and consequently PQ is inclined 
 to MN at the same angle that AC is toOwv; and since MN, 
 NP, are the co-ordinates of P in the plane QN, 
 
 PN=A4A.MN + QM. 
 
6 
 
 But, Q being a point in the line BC, QM = B.MO+ OC, 
 . PN=A.MN+B.MO+0C; 
 or, if v, y, s denote the co-ordinates of P, 
 s=Av+ By +e, 
 
 the equation to the plane. 
 
 Cor. The lines in which a plane intersects the co- 
 ordinate planes are called its traces on those planes; thus 
 AC, BC, are two of the traces of the plane BCA, the third 
 being the line in which, if prolonged, it would intersect the 
 plane of wy. Hence, we perceive the meaning of the constants 
 in the equation to a plane, 
 
 s=Av+Byt+e; 
 
 for A, B, are the tangents of the angles at which the traces on 
 #&,Yy%, are respectively inclined to the axes of w and y pro- 
 duced in the positive directions; and ¢ is the portion of the 
 axis of x produced in the positive direction, intercepted 
 between the plane and the origin. 
 
 It is important to observe, that the foregoing method 
 of finding the equation to a plane, applies equally to the case 
 where the co-ordinates are oblique, and leads to a result of 
 the same form; so that whether a plane be referred to rect- 
 angular or oblique axes, its equation may be represented by 
 x= Awxv+ By+c; but in the latter case, A will signify the 
 ratio of the sines of the angles at which the trace on xa is 
 inclined to the axes of w# and x; and B the ratio of the sines 
 of the angles at which the trace on yz is inclined to the axes 
 of yand x. Hence, all results which involve no other as- 
 sumption than s = 47+ By +e, for the form of the equation 
 to a plane, will be equally true for oblique and rectangular 
 co-ordinates. 
 
 8. To investigate the equation to a plane under the form 
 
 ve 41 x 
 weet AE as) 
 a C 
 
7 
 
 Let AC (fig. 4) in the plane of za be the straight line to 
 which the generating straight line is always parallel, BC in 
 the plane of yz that along which it moves; Ga point in the 
 generating line DE in any position, ON=a, NH=y, HG =z, 
 
 its co-ordinates. 
 
 Draw DF, FE parallel to OC, OA, and let OA, OB, OC, 
 which are called the intercepts of the co-ordinate axes be re- 
 spectively denoted by a, b,c. 
 
 Then since G is a point in DE, 
 
 & wv 
 
 p+ at erin 
 DF'F (1) 
 but ere Se aie ye 
 Cc b a b 
 5 ae DF 
 therefore, multiplying the first term of (1) by ——, the second 
 c 
 
 FE 
 , and the second member by the equal quantity 
 
 term by 
 a 
 
 1 = and transposing, we get 
 
 —$- 4 5=] 
 hig tal ok ; 
 a relation holding between the co-ordinates of any point in the 
 generating line in any position, and consequently the equation 
 to the plane generated, and expressed by the intercepts of the 
 co-ordinate axes produced in the positive directions to meet the 
 plane. 
 
 Cor. It is easily seen that this result might have been 
 deduced from Art. 7. For if we take s = dv + By +c to be 
 the equation to the plane ABC (fig. 4), and in it make z = 0, 
 y=0, «=a for the point 4, we get 0= Aa+c; and making 
 s=0, w=0, y=b for the point B, we get 0= Bb+c; there- 
 fore, substituting for A and B their values, the equation to the 
 plane becomes, as above, 
 
 a 
 
8 
 
 9. To investigate the equation to a plane under the 
 form wcosa+y cos + x cosy =p. 
 
 Let OQ =p (fig. 4) be the perpendicular from the origin 
 upon the plane 4BC, making angles a, (3, y with the axes of 
 &, Y, ¥, respectively ; join AQ, then, OQA being a right angle, 
 
 OA = OQ sec AOQ, or a = pseca; 
 
 similarly, 6b = psec3, c =p secryy; consequently, by substi- 
 tuting in the equation 
 
 oy 
 
 cD 
 
 we get 
 xcosa+ycos 3 + x cosy =p, 
 
 the equation to a plane in terms of the perpendicular let fall 
 upon it from the origin, and the angles which that perpen- 
 dicular makes with the co-ordinate axes produced in the 
 positive directions. 
 
 It will be observed that neither in this article, nor the 
 preceding, are the co-ordinates required to be rectangular ; 
 only, in the case of rectangular axes we must have (Art. 5) 
 
 cos’ a + cos’ 3 + cos’ y = 1; 
 
 but in the case of oblique axes the cosines, which fix the 
 position of the normal to the plane, will be subject to a 
 different condition. 
 
 Cor. If with the above equation we compare the general 
 form of the equation to a plane 
 
 Ax+ By+ Cz=D, 
 
 im) b) 
 p cosa cosp cosy 
 
 which shew that the constant term bears the same ratio to the 
 perpendicular on the plane, that the coefficient of each variable 
 
9 
 
 does to the cosine of the angle which its axis makes with the 
 perpendicular. 
 
 Hence the perpendicular on a plane from the origin 
 
 constant term 
 
 s/ sum of squares of coefficients of variables 
 
 likewise the square of its reciprocal equals the sum of the 
 squares of the reciprocals of the three intercepts of the co- 
 ordinate axes. 
 
 10. The particular cases of the equation to a plane, 
 when it involves only one, or only two of the variables, 
 require a separate notice. 
 
 Thus the equation #2 =a, since it belongs to all points 
 whose distance from the plane of yz, measured parallel to the 
 axis of #, is a, represents an indefinite plane parallel to yz. 
 Similarly, the equation w = 0, characterizes all points situated 
 in the plane of yz, or is the equation to that plane; and 
 
 y=0, x=0, are respectively the equations to the planes of 
 SL, LY. 
 
 Again, the equation — + 5 =1, 
 a 
 
 represents not only the line 4B (fig. 6) in the plane of ay, 
 but all points situated in the plane RA drawn through 4B 
 parallel to the axis of x ; for any one of these points P, what- 
 ever be the value of PM or x, will have the same wv and y 
 as its projection M, and therefore its co-ordinates will satisfy 
 the equation 
 
 which does not involve x. Similarly, the equations # = mz+a, 
 y =nx +, represent planes parallel to the axes of y and 2; 
 i.e. each equation represents a plane parallel to that axis 
 whose co-ordinate it does not involve. 
 
 This also appears from the equation z = da+ By+c; 
 for if the plane be parallel to the axis of y, then its trace on 
 
10 
 
 yz will be parallel to the axis of y, and therefore B =0, or 
 x = Awv+cis its equation, the same as the equation to its 
 trace on yz; similarly, if the plane be parallel to the axis of 
 x, 4=0. In order to deduce from it the equation to a plane 
 parallel to the axis of x, the equation requires to be reduced, 
 as in Cor. Art. 8, to the symmetrical form 
 
 now let ¢ be infinite, then the plane becomes parallel to the 
 
 axis of x, and its equation 1s — + : =1, the same as that 
 a 
 
 to its trace AB. If the plane pass through the origin, 
 
 c= 0, and its equation is 7 = 4v+ By. 
 
 11. Every equation of the first order involving three 
 variables is the equation to a plane. 
 
 Let the equation be reduced to the form = Awv+ By + ¢;3 
 and to find the trace of the surface which it represents on yx 
 letv =0; .. x = By+ ce, which is the equation to a straight 
 line ; let this be CB (fig. 5), Next to determine the sections 
 made by planes parallel to va, let y=b,0’, &c. «1. x= Av+Bb+t+e, 
 z= Axv+ BY +c, &c. which represent straight lines QP, OP’, 
 &c. parallel to one another. They likewise all intersect the 
 trace CB; for making w«=0, we have Bb+c, Bb’ +c, &e. 
 for the values of QM, Q’M’', &c. and these are the ordinates 
 of CB corresponding to y=b, y= 6, &c. Therefore the locus 
 of the proposed equation is a system of parallel straight lines 
 all passing through a fixed line; i.e. it is a plane, 
 
 12. The equation x= Ar+ By +e, 
 where A, B denote numbers, and ¢ a line, has the defect of 
 not being symmetrical; to compensate for which, it contains 
 only three constants, each having an obvious signification so as 
 to enable us readily to interpret any results to which its use 
 may lead; and it furnishes the analytical expressions of the 
 
 more useful relations of lines and planes to one another, under 
 — 
 
11 
 
 forms the simplest possible ; and it is readily adapted to planes 
 in all particular positions, with the single exception of a plane 
 parallel to the axis of x. But it may often be advantageously 
 replaced by the symmetrical form expressed by the intercepts 
 of the co-ordinate axes (which however cannot represent a 
 plane passing through the origin) 
 
 or by the equation depending on the magnitude and position 
 of the perpendicular from the origin upon the plane, which 
 may be written 
 
 lv+my+ns=D, 
 
 if we take 1, m, m to denote the cosines of the angles which 
 the perpendicular p forms with the axes of a, y, #3; and which 
 may represent a plane in every position. ‘The above three 
 forms of the equation to a plane, as has been observed, hold 
 equally for rectangular and oblique co-ordinates. 
 
 13. To find the equations to a straight line in space. 
 
 A straight line is the intersection of two planes, and con- 
 sequently is given when the equations to any two planes 
 which contain it are given. 
 
 Among the various planes which may be drawn through 
 a given straight line, those are employed, for the sake of 
 simplicity, to determine it, which are parallel to the co- 
 ordinate axes, since their equations will involve only two of 
 the variables. 
 
 From all the points of the line MM” in space (fig. 6, bis) 
 let perpendiculars MC, M’'C’, &c., be drawn to the plane of 
 xy, meeting it in the points C, C’, &c.; the assemblage of 
 these points is called the projection of MM” on this plane. 
 This projection will manifestly be a straight line, since all the 
 perpendiculars are situatad in the same plane, which is parallel 
 to Ox, and is called the projecting plane of MM”. Similarly, 
 if we project IM” upon the other co-ordinate planes by per- 
 pendiculars upon them from its various points, we shall have 
 
12 
 
 the three projections 44’, BB’, CC’, of which any two are 
 sufficient to determine MM’. For suppose Ad’ and BB’ are 
 given; if through these we draw planes respectively perpen- 
 dicular to ga, yx, each of them must contain the line MM, 
 and will therefore fix its position in space by their intersection. 
 But the positions of the projections NA, QB in the planes 
 za, yx, will be determined by their equations 7 =mz +a, 
 y=nx-+b6, in which m and m denote the tangents of the 
 angles at which they are respectively inclined to the axis of 
 zg, and a=ON, b=OQ. Hence the system of equations 
 
 G=IMS+a, y=nsv4+, 
 
 which belong to the projections of the line, or rather to its 
 projecting planes, will completely determine it. 
 
 Cor. 1. Any values of x, y, x which satisfy the above 
 equations are co-ordinates of some point in the line; for sup- 
 pose wv’, y’, x’, were such values, and let M’ be the point in 
 MP for which x = 2’; then for each of its projections 4’, B’, 
 g =’, and therefore a’ = ON’, y’ = OQ’; that is, if x’ = C'M, 
 then a and y’, provided they be determined from the above 
 equations, will equal ON’, N’C’. Hence the equations =mz+a, 
 y = nx +b, may be called the equations to the line MP itself. 
 
 Cor. 2. Corresponding to the value M’C’ of x, the 
 values of a and y in the simultaneous equations 7 = mz +a, 
 y=nz+b, are ON’, NC’, the co-ordinates of the point C’ 
 of the projection PC; and therefore the equation to the 
 third projection may always be deduced from the two others 
 by eliminating x, which gives 
 
 w-a y-b 
 = » or nv —- my = na—mb. 
 
 m 
 
 14. In the same manner it might be shewn, that if the 
 axes of the co-ordinates were oblique, the equations to a 
 straight line would still be of the form 
 
 L=ME+a, Y=nsv4+db; 
 
 the only difference would be that the projections would be 
 made by planes parallel to the axes of the co-ordinates; and 
 
13 
 
 m would denote the ratio of the sines of the angles which the 
 projection on x# makes with the axes of x and w respectively ; 
 and similarly for . 
 
 15. The point in which a line meets any one of the 
 co-ordinate planes, is called its trace upon the plane. The 
 positions of these points are determined by putting, in the 
 equations to the line, a, y, x, successively = 0; thus the co- 
 ordinates of the traces. of the line whose equations are 
 
 v-a y-b 
 
 L=ME+a, y=nz4+b, = —_ 
 m n 
 
 3 
 
 upon the planes of yz, zx, wy, are respectively 
 
 a b 
 a — = 
 m n rie 
 =b 
 yaa 48| re A 
 
 16. Asin the case of a plane, it is sometimes attended 
 with great convenience to have the equations to a straight line 
 expressed in symmetrical forms. 
 
 If a, 2, y be the co-ordinates of a given point in a straight 
 line, and w, y, x those of another point in the line at a distance 
 r from the former, and /, m, n, be the cosines of the angles at 
 which the line is inclined to the axes of vw, y, =, we have 
 
 (Cor. Art. 6) 
 
 and as these relations subsist amongst the co-ordinates of every 
 point in the straight line, they are its equations. 
 
 This will still be true when the axes are oblique, if J, m, 
 m are the ratios of the projections of a line on the co-ordinate 
 axes, to the line itself. 
 
 Problems on the plane and straight line. 
 
 These principles being laid down, we proceed to the reso- 
 lution of several Problems relative to the plane and straight 
 
14 
 
 line, the results of which are of great use. As the equation 
 to a plane contains three disposable constants, a plane may be 
 drawn so as to fulfil various conditions; as, for instance, to 
 pass through three given points; to pass through a given point 
 and be parallel to a plane, or to each of two given straight 
 lines; or to pass through a given point and be perpendicular 
 to a given straight line. Similarly, a straight line, since its 
 two equations contain four disposable constants, may be drawn 
 so as to satisfy various conditions. 
 
 17. To express that a plane passes through a given point. 
 
 Let zs = da + By +c, be the equation to a plane passing 
 through a point av’y’s'; then the equation to the plane must 
 be satisfied by the co-ordinates of the point, .. s’= Aa’+ By'+c¢; 
 therefore, subtracting this equation from the former to elimi- 
 nate c, we have 
 
 w-2'= A(w-a') + BYy-y), 
 
 in which A and B remain undetermined, so that the plane 
 may still be made to satisfy two other conditions. 
 
 Similarly, if a, 8, yy be the angles which the perpendicular 
 on the plane forms with co-ordinate axes, and the plane pass 
 through a point w’y’s’, its equation will be 
 
 (w — aw’) cosa + (y—y') cos B + (x — 3’) cosy = 0. 
 
 18. To find the conditions that two planes, whose 
 equations are given, may be parallel to one another. 
 
 Let x= dv + Byt+c, x= A'x+ By +c’, be the equa- 
 tions to two planes; then if they be parallel to one another, 
 their common sections with any plane which cuts them are 
 parallel; therefore their traces on the co-ordinate planes are 
 parallel. But, making y = 0, the traces on the plane of sa# 
 have for their equations s = dv+c, s = Aa +c, which will 
 be parallel if 4 = 4’; similarly, since the traces on yz are 
 parallel, B = B’. Hence the conditions of parallelism of two 
 planes are, that the coefficients of w and y in their equations 
 are respectively equal. 
 
15 
 
 Cor. The equation to a plane drawn through a point 
 a yx’, parallel to a given plane x = 4’7+ By +e, is 
 
 s—-x =A (w-wv)+B(y-y). 
 
 19. To find the equation to a plane which shall pass 
 through three given points. 
 
 Let 2, 91215 %oY2%2y V3Y3%3, be the co-ordinates of the 
 given points, and s = dv + By +c the equation to the plane ; 
 then the co-ordinates of the points must satisfy it ; 
 
 2 =Av,+ By, +c, 
 %= Av, + By. +, 
 %3 = Axv;+ By; + ¢, 
 which are the three equations for finding A, B,c. Multiply 
 
 the second and third respectively by indeterminate coefficients 
 t, w, and add them to the first ; 
 
 °, 8 + t%. + U%3 = A (a, + FH, + UH5)...... Clie 
 provided y, +¢y,+ uy,;=0, 1+t+u=0O. 
 These two latter equations give 
 YWi-Ystt(Y2—Ys)=% Yr—Y2t U(Ys — Y) = 0, 
 and so determine ¢ and zw; and substituting these values in 
 
 equation (1), we find A; and then B may be obtained by 
 
 interchanging a,,7, and Y,Y2y3; so that 
 #—#,=A(w-4,)+ Biy-y), 
 the equation to the plane, will be completely determined. 
 
 20. ‘To determine the line of intersection of two planes 
 whose equations are given. 
 
 Let the equations to the planes be 
 s=Av+By+e, x=Av+By+e. 
 
 _ Then the line of intersection is, as we know, sufficiently de- 
 _ termined by these two equations taken simultaneously; that is, 
 by supposing w, y, x to receive only such values as satisfy 
 
16 
 
 both at the same time. If, for instance, we substitute any 
 series of values for x, we shall have for each of these values 
 two equations to find the corresponding values of w and y; 
 and thus we may determine as many points in the line as we 
 please. But the line of intersection is more conveniently de- 
 termined by its projections on the co-ordinate planes. To find 
 these projections, let a’y's’ denote the co-ordinates of any point 
 in the intersection, then 2’, x’, are also the co-ordinates of the 
 corresponding point of the projection on za; 
 
 . &=Aa't+ Byte, # = Aa t+ By +c; 
 therefore, eliminating y’, 
 (B’ — B)s' = (AB — A'B)a' + Be - Be; 
 
 which is the relation between the co-ordinates of any point in 
 the projection on zw. Hence, suppressing the accents, 
 
 (B' - B)z = (AB — AB) & + Be — Be’, and, similarly, 
 (A’ — A)x = (BA'—- B’A)y + A’c - AC,~z 
 are the equations to the line of intersection of the planes; 
 which, as we see, result from separately eliminating y and x 
 
 from the equations to the planes. The projection on wy has 
 evidently for its equation 
 
 (A’— A)v + (B’- B)y+c-—c=0. 
 
 21. To find the equations to a straight line which shall 
 pass through two given points. 
 
 VE OF OP 
 
 Let a’y 2’, v’y”z”, be the co-ordinates of the given points ; 
 
 L=MS+a, y=nsx +b, the equations to the line, 
 
 which must be satisfied by the co-ordinates of the points ; 
 
 * v0 =ms +a, yons + D....., (1), 
 4} ” ” 4? 
 C=HamMs+a, y=ns +d, 
 the four equations to determine the four constants. First, for 
 the values of m and n, we have 
 
 uv’ med a” =m (x' rl 2"), y wr y” = n (2 2") ; 
 
17 
 
 also subtracting equations (1) from the assumed equations in 
 order to eliminate a and J, 
 
 v-v =m(x-2), y-y =n(e-%); 
 
 therefore, substituting for m and n, the required equations are 
 av —- a” y y” 
 , ar ’ t “hal , 
 Gig Bilan Ae = *); Sas Unc a ten Pig S ds 
 
 Cor. 1. It is easily seen that if we take abe, a’b’c’, for 
 the given points, and call their distance d, and the distance 
 of abe from wyz, r, the equations to the straight line may 
 be written 
 
 @-a y-b B8-C¢ 
 
 Qls 
 
 eed ah Oe ah Oo 
 Cor. 2. If the line is required to pass through only one 
 point a’y’s’, its equations will be 
 e-al=m(z-2), y-y=n(s-2), 
 
 where m and m remain undetermined, as might have been 
 foreseen ; for this condition may be satisfied by an unlimited 
 number of lines. If, besides, it is to be parallel to a known 
 line of which the equations are 
 
 c=mzta, yans4+0, 
 
 since the projecting planes (uc. x1. 15) and therefore the pro- 
 jections of these lines must be parallel, we have m = m', n =n’; 
 therefore the required equations are 
 
 e-xv=m(e-2%) y-ya=an (x —-2). 
 
 22. To determine the point of intersection of two given 
 lines. 
 
 Let their equations be 
 L=Ms+a c=omz+a 
 eae eapeaety 
 For any given value of z, the values of w and y furnished by 
 these two systems of equations will in general be different ; 
 
 2 
 
18 
 
 but if s be the ordinate of their point of intersection, the cor- 
 responding values of w and y furnished by them, will be the 
 same. Therefore the co-ordinates of the point of intersection 
 are the values of wv, y, %, which simultaneously satisfy the 
 above four equations ; and as the number of equations exceeds 
 the number of quantities to be determined, there will be an 
 equation of condition, without which the problem would be 
 impossible; since two straight lines do not, in general, intersect. 
 Hence, eliminating # and y separately, 
 
 O=(m—-m)zx+a-a, 0=(n-n')x+b—-8; 
 and equating these two values of z, 
 a-a 6b-D' 
 
 it, 
 m—m rm—-nN 
 
 3 
 
 an equation which must be made identically true by the con- 
 
 stants entering into the equations to the two lines, in order 
 
 that they may intersect. The two remaining co-ordinates of 
 
 the point of intersection, will be found by substituting the 
 value of 
 
 a-a 6-0 
 
 3 = = = eit 
 
 ? Le 9 
 m—-m wu— 7 
 
 in either of the above systems of equations. 
 
 Cor. When m=m’, and n=7’, the equation of con- 
 dition is satisfied, and yet the lines do not intersect, being 
 parallel; therefore the above condition will not ensure the 
 intersection of the lines, without the restriction that the value 
 of zs is not to be infinite. Taken without any restriction, 
 it expresses that the lines are in the same plane and may ~ 
 therefore intersect; without deciding at what distance that — 
 intersection will happen. 
 
 23. To find the conditions in order that a straight 
 line and plane whose equations are given, may coincide, or 
 be parallel to one another. 
 
 Let s = da + By +, be the equation to the plane, 
 
 V=MS+a, y=nz+), 
 
 a 
 
19 
 
 the equations to a line coinciding with it. Since every point 
 of the line is in the plane, if x be the ordinate of any point, 
 and ... mz +a, nz +, the values of the other co-ordinates, 
 these values must satisfy the equation to the plane; 
 
 . O=(4m+ Bn-1)24+Aat+Bb+e; 
 and since this equation must be true for all values of z, 
 “~. Am+Bn-1=0, Aa+Bb+c=0, 
 which are the required conditions. 
 If the plane and line are parallel, then a plane and 
 line drawn through the origin respectively parallel to them, 
 
 coincide; therefore the above equations must be satisfied when 
 we suppose ¢, b, a, to vanish; hence the remaining equation 
 
 Am+ Bn-1=0, 
 
 expresses the condition in order that a line and plane may be 
 parallel to one another. 
 
 24, If we employ the symmetrical forms of the equations 
 to the straight line and plane, then in order that the straight 
 line 
 
 may be parallel to the plane da + By + Cz=D, we must have 
 lA+mB+nC=0; 
 
 and if it coincide with it, the further condition is 
 Aa+Bb+Cc=D. 
 
 It will be observed that the results of the preceding 
 Articles from (17) inclusive, are also true, when the axes are 
 oblique ; since the only assumptions made, are the forms of 
 the equations to the plane and straight line; but in that case, 
 the constants will have different significations. 
 
 25. ‘To find the conditions that a straight line and plane 
 may be at right angles to one another. 
 2—2 
 
20 
 
 Let the equations to the straight line and plane be 
 
 S Gnt hg and te 4 wor Bane 
 y=ns + b, 
 then the equations to the straight line and plane respectively 
 parallel to them through the origin are 
 (1). s=Auv+By (2). 
 
 eC=MsZ 
 
 y= ns 
 
 Now let OA (fig. 8) be the straight line (1), then since it 
 is perpendicular to the plane (2), it is perpendicular to every 
 straight line meeting it in that plane, and consequently is 
 perpendicular to the trace OB of (2) on the plane of #z whose 
 equations are 
 
 z= Ax, y=0 (38). 
 
 In straight lines (1) and (3), take two points, 4, B whose 
 co-ordinates are respectively a’, y’, 3’; #”, 0, 2°; and join 
 those points; then by equating the two values of the hypo- 
 thenuse of the right-angled triangle 4OB so formed, we get 
 AB’ = AO’ + OB’, or 
 
 (a — a") + yy? + (eX — 2" Poa ey? + ev? te? ee”; 
 U0 +22 =0, 
 or since a’ = mz’, 2" = Av”, m+A=0. 
 Similarly, for the trace on yz, we get 2+ B=0; hence 
 the required conditions are 
 m+A=0 n+B=0. 
 
 Cor. 1. Ifit were required to find the conditions for the 
 line and plane represented by the equations 
 
 DONT oy oT hesom CNIS Say 
 l a ee 
 
 5) Auv+ By+Cz=D 
 
 being perpendicular to one another, since the line is parallel to 
 
 the normal to the plane, those lines must form equal angles — 
 
 with the co-ordinate axes; but the cosines of the angles which 
 
 the line forms with the axes of x, y, x, are proportional to 5 
 
 l,m, 2; and the cosines of the angles which the normal forms 
 
21 
 
 with the same axes are proportional to 4, B, C (Arts. 9 and 16) ; 
 
 Lia ed 
 
 Cor. 2. Hence we can find the equations to a straight 
 line which shall pass through a given point, and be perpen- 
 dicular to a given plane. 
 
 Let a’, y’, x’, be the co-ordinates of the given point, 
 x = Aw + By +c the equation to the plane; 
 then » — a =m(z —), y-y =n(z% —2’) are the forms of 
 
 the required equations (Cor. 2. Art. 21); and since the line is 
 perpendicular to the plane, 
 
 m=-— A, n=-—B, 
 - 2—-v' + A(z—2) =0, y-y + B(z -2%) =0, 
 
 are the required equations. 
 
 26. Since the trace of the given plane on vw has for its 
 equation xs = Aw + ¢, and the equation to the projection of the 
 
 . e . av a 
 given line on sw may be written x =— — —, the result 
 mm 
 
 1 
 A+m=0, or Ax —+12=0, shews that the trace and the 
 m 
 
 projection on sa, and consequently on any plane whatever, are 
 perpendicular to one another. The geometrical interpretation 
 of the result of Art. 25 consequently is, that when a straight 
 line is perpendicular to a plane, the projection of the line and 
 the trace of the plane on each of two co-ordinate planes, are 
 perpendicular to one another. ‘The only exception which this 
 assertion is liable to, is when the plane is parallel to one of 
 the co-ordinate axes, as the axis of x; for in that case the 
 two projecting planes become coincident, and a line CQ (fig. 6) 
 may have two of its projections CS, CR perpendicular to AS, 
 BR the parallel traces on x@ and yz, without being perpen- 
 dicular to the plane AR. 
 
 b 
 Suppose that y = —— w + b was the equation to the plane; 
 : a 
 
 then every line perpendicular to it would be in a plane parallel 
 
22 
 
 to wy, and therefore have « =c for one of its equations, 
 a ° : Ale 
 and y = ,et b’ for the other, since its projection on vy must 
 
 be perpendicular to the trace on that plane. 
 
 27. If through CO (fig. 7) the axis of z, a plane COQ 
 be drawn perpendicular to AB, the trace of the plane ABC 
 on wy, and OP be drawn perpendicular to CQ, it is evident 
 that OP is perpendicular to plane ACB and z CQO is the 
 inclination of the plane ABC to the plane of wy, and is equal 
 
 to the 2 COP, which OP makes with the axis of x, each being © 
 
 equal to the complement of 2ZPOQ; in other words, the 
 
 angle between two planes is equal to the angle between two — 
 
 lines respectively perpendicular to them. If the perpendiculars 
 should not meet, then by the angle between them, and gene- 
 rally by the angle between any two lines that do not intersect, 
 we understand the angle formed by two lines drawn through 
 any the same point respectively parallel to them. 
 
 28. Having given the equations to a straight line, to 
 find the angles which a line parallel to it through the origin, 
 makes with the axes. 
 
 Let v= mz+a, y=nz +b, be the equations to the given 
 
 line. Through the origin draw a line OA (fig. 8) parallel to — 
 
 it; then the equations to OA are 
 
 V=EMZ YHNS, 
 
 Let a, 3, ry, denote the angles which OA makes with the © 
 
 axes of x, y, #3 also let OA = 1, and let 2’, y’, ', be the co- 
 
 ordinates of the point A, then (Art. 5) a =cosa, y’ = cos P, — 
 
 % = cosy; also since the point is in OA, 
 Ud , , ’ 
 Cz MB 5 Y = 72S. 
 But OA? = v7? +y? +2", or 1 = (m? +n? +1) 2”, 
 ; 1 1 
 aa ES i ee COS SS 
 Ji +m +n y /1 +m +n? 
 n m 
 
 cos 3 = ——————., COs | 
 /1+m + n° Y/1l+m +n* 
 
23 
 
 The radical in these expressions admits of a double sien 
 + or —; but as it must be taken + in all, or — in all, we 
 have only two systems of values for a, 6, yy; viz., the angles 
 which OA makes with Ow, Oy, Ox; or their supplements, 
 which AO produced or OA’ makes with the same lines. It 
 is usual to take the positive sign; then cosy is +, and 
 therefore -y acute; and the angles determined are those which 
 the part of the line above the plane of wy makes with the co- 
 ordinate axes produced in the positive directions. 
 
 Cor. Conversely these equations give the values of the 
 coefficients m and nN, 
 
 cos a cos (3 
 
 m = | 7 ae e 
 cos y cos y 
 
 Hence the equations to a line passing through a given point 
 (v’, y’, 2’) and making angles a, 3, y with the axes, are, as 
 we have already found at Art. 16, 
 
 Uy , U 
 0-@ Y-Y 8-8 
 
 cosa cos B cos 
 
 29. To find the angles of inclination of a plane whose 
 equation is given, to the co-ordinate planes. 
 
 Let s = Av + By +c be the equation to the plane ; 
 
 then v=-— Av, y= — Ba, 
 
 are the equations to a line through the origin perpendicular 
 
 to it; and if a, 3, y denote the angles which this line makes 
 
 with the axes of a, y, z, then (Art. 27) a, 8, y are also the 
 
 angles at which the plane is inclined to the co-ordinate planes 
 of yx, xv, wy respectively ; and by the preceding article we 
 have 
 
 -—A -B 
 
 cos @ = cos: BP = —— 
 /1+4+ A? + B? /1+ 47+ B 
 
24 
 
 30. Having given the equations to two straight lines, to 
 find the angle which two lines parallel to them drawn through 
 the origin include. 
 
 Let 
 
 , 
 
 y=ans+ 
 
 a. v=ms+a’ 
 ‘ Y ? 
 
 lines to which OA, OB, (fig. 8) are respectively parallel ; then 
 L=ms, y=nsz are the equations to OA, and r=m sz, 
 y=n's, those to OB. Take OA, OB, each=1; and let 
 the co-ordinates of A and B be denoted respectively by 
 L,Y, %3 @,y,2%3 join AB, and lett AB=d, 2 AOB=8. 
 Then from the triangle AOB, ~ 
 
 be the equations to the 
 ns +b 
 
 124+1?-2cos0=ad’ =(w — a’)? 4+ (y—-y')? + (x — 2’) 
 or, expanding, and reducing by the relations | 
 e+y+e2?a1, w+ y? 4+ 2% = 1, 
 cosO0 = wa’ +yy' + 22". 
 But since 4 and B are situated respectively in OA, OB, 
 L=Ms, y=ne; 2 =m, yY =n's; 
 
 1 
 
 J 1 +m +n? 
 
 . Ll=(m'? +n’? +1)2", or v= 
 
 1 
 Sia larl ve ver es ap pe 
 ae /1+m? +n?’ 
 
 , , 
 . cos@ = (mm'4 nn’ +1) 23 = nT DE OES : 
 Si tm +n /1 +m? +n’ 
 
 This expression, on account of the two radicals it con- 
 tains, will furnish for 6, four values equal two and two, 
 corresponding to the four angles formed by the indefinite lines 
 
 Ad’, BB’. It is usual to take the denominator positive, — 
 
 i. e. to take the radicals with the same sign, so that the 
 points 4, B, are both above or both below the plane of wy; 
 and the angle determined is AOB or its opposite 4’OB, 
 contained between the two portions of the lines which form 
 
25 
 
 each an acute angle with Ox or Ox’; this angle, moreover, 
 will be acute or obtuse according as 1 + mm’ + nm’ is positive 
 or negative. 
 
 Cor. By means of the formula sin = rv gee (cos)? we may 
 shew that 
 /(m — m')? + (n— n+ (mn' — m'n)? 
 
 sin 9 = fa 
 Si tm +n? /1 +m? +n? 
 
 31. The angle between two lines may also be ex- 
 pressed in terms of the angles which each makes with the 
 co-ordinate axes. 
 
 Let OA make with the axes of w, y, x the angles 
 a, 8, y; and OB make with them the angles a’, 8’, y’; 
 then, proceeding as in the last Art., we may shew that 
 
 cos 0 = wa’ + yy + 22". 
 But vx=cosa, y=cos, % = cosy; 
 
 , , ? ‘ , : , 
 
 @ =cosa, y =cosf, % =cosy ; 
 
 *. cos@ = cosacosu + cos 3 cos 3’ + cos y cosy’. 
 32. Hence the condition in order that two lines may be 
 perpendicular to one another is cos @ = 0, or 
 mm +nn +1=0; 
 
 at which we may arrive immediately by observing that the 
 equation to a plane through the origin perpendicular to the 
 first line is 7 + maw +ny=0, which must necessarily contain 
 a line through the same point parallel to the second whose 
 equations are w= m'z, y=; 
 
 “St mms+nn'z=0, or mm+nn'+1=0. 
 This condition expressed in another form, is 
 
 cos a cos a’ + cos $ cos B’ + cosy cosy’ = 0. 
 
26 
 
 Also, in order that the lines may be parallel to one another, 
 we must have 
 
 sinO=0, or (m—m’)? + (n—17')?+ (mn' — m'n)’= 0, 
 
 which can only be satisfied by m=m’, n=n’'; this agrees 
 with Cor. 2, Art. 21. 
 
 ee ———— 
 
 33. To find the angle of inclination of two planes whose ~ 
 
 equations are given. 
 
 Let s=Av+ Byt+e, s= Ax + By +c, be the equa- 
 tions to the two planes. Then the equations to two lines 
 
 respectively perpendicular to them through the origin are 
 (Art. 25) 
 
 e=-Ax, y= —-Bz; w= -— A's, y= —- B82; 
 and (Art. 27) the angle of inclination, 0, of the two planes 
 
 is equal to the angle between these two perpendiculars ; 
 therefore (Art. 30) 
 
 9 1+ 44+ BB 
 COs = ee ee ee 
 af/1+ + B/1 +A 4 BE 
 
 the double sign belonging to the acute and obtuse angles 
 between the two indefinite planes. 
 
 34. The angle of inclination of two planes may also 
 be expressed in terms of the angles at which each is inclined 
 to the co-ordinate planes. 
 
 For if a, B,y3; a, By’; denote the angles at which 
 the planes are inclined to the co-ordinate planes of ys, za, 
 xy, these are also the angles which the perpendiculars on 
 them from the origin respectively make with the axes of 
 #2, y, #3; therefore (Art. 31) 
 
 cos 9 = cosa cos a’ + cos 2 cos (3 + cosy cos ¥’, 
 35. Hence the condition that two planes may be per- 
 
 pendicular to one another (since in that case cos @ = 0) may 
 be expressed in the two following ways: 
 
27 
 1 a AALS BB’ = 0, 
 cos a cosa’ + cos 3 cos 3’ + cosy cosy’ = 0. 
 
 We may arrive at the former of these results immediately 
 by observing that v= — dz, y = — Bx, are the equations to 
 
 a perpendicular through the origin to the first plane which 
 _ must coincide with a plane through the origin parallel to the 
 second plane whose equation is s = 4’x + By; 
 
 * AA t BBA Ie 0: 
 
 If the second plane be parallel to the axis of x so 
 that its equation is y=atan@+ 6, the condition becomes 
 
 B= Atan ¢. 
 
 36. To find the angle between a straight line and a 
 plane, of which the equations are given. 
 
 The angle, 0, here meant, is the angle between the 
 line and its projection upon the plane, and is therefore equal 
 to the complement of the angle made by the line with a 
 perpendicular to the plane. 
 
 Let s = dv + By +c be the equation to the plane; then 
 the equations to a line through the origin perpendicular to 
 it are 
 
 e=—-Azx, y= — Bz. 
 
 Let v=ms+a, y=nx+b, be the equations to the 
 
 : : uu : : 
 given line; then since aoa is the angle between it and 
 
 ~ 
 
 the perpendicular to the plane, 
 mute ae 1—-Am— Bn 
 cos (= _ ) or sing = /1 tne + 8 Viepeee Be ° 
 
 37. To find the length and the co-ordinates of the 
 extremity of the perpendicular from a point upon a plane. 
 
 Let s=Axv+By+ec be the equation to the plane, 
 x, y’, x the co-ordinates of the point from which a perpen- 
 
28 
 
 dicular is to be dropped upon it; then the equations to the 
 
 line passing through the point and perpendicular to the plane | 
 
 are (Art. 25, Cor. 2) 
 e-av =—A(xsx-2), y-y=-B(s-2); 
 
 and combining these with the equation to the plane, which 
 may be written 
 
 s-s'=A(ea—-v)+B(y-y) -(' -Aa' - By -0), 
 
 in order to deduce the co-ordinates XY, Y, Z of the point 
 where the perpendicular meets the plane, and putting 
 
 x — Aa’ — By’ -—c=P, we get 
 Z—-2# =-A(Z-3') -B (2-2) -P, 
 
 075 — is a 
 
 AP ; BP 
 
 Xe pete Ee igs == ——________ : 
 = ieee ee J 1 + A424 Be” 
 
 values from which it is easy to deduce X, Y, Z, the co- — 
 
 ordinates of the foot of the perpendicular. 
 
 If we denote by D the length of the perpendicular inter- — 
 
 cepted between the point and the given plane, we have 
 P* 
 1+ A? + B’ 
 2’ — Aw’ — By -c 
 £4/14 4° +B” 
 the radical being taken with that sign which makes the whole 
 expression positive. 
 
 D?=(Z- x)? +(¥ -y)’+ (X-2)’ = 
 
 . D= 
 
 Cor. If we use the symmetrical form of the equation to 
 a plane 
 Av+ By+Czs=D, 
 D 
 / A? + B+ C?’ 
 
 and a plane parallel to it through (abc) has for equation 
 Axvx+ By+ Cz =Aa+ Bb+ Ce, 
 
 the perpendicular upon it from the origin = 
 
29 
 
 the perpendicular upon which from the origin equals 
 
 Aa+ Bb+Cc 
 Aa+Bb+Cc~D 
 which is the perpendicular from (abe) upon the proposed 
 plane. : 
 
 the difference of these perpendiculars = 
 
 ? 
 
 38. To find the length and the co-ordinates of the 
 extremity of the perpendicular from a point upon a line. 
 Let the equations to the given straight line be 
 V=ME+a, y=nsth, 
 
 and a’, y’, x’ the co-ordinates of the given point from which a 
 perpendicular is to be dropped upon it; then the equation to 
 a plane passing through (a’, y’, ’) and perpendicular to the 
 given straight line is (Art. 25) 
 
 #—s'=—m(x-a)-n(y-y); 
 and combining this with the equations to the line in order to 
 
 get the co-ordinates X, Y, Z, of their point of intersection, we 
 have, putting 
 
 x +m(a —a)+n(y -b) =P, 
 Zax’ —m(mZ+a)+ma—n(nZ +b) + ny’; 
 x +m (uv —a)+n(y —b)_ 
 
 » 2 
 1 +m? +7? H 
 Reelily sno. i'd.) 
 1+m’ +n? 
 m P 
 aC nr cae oO) 
 m n 
 FNC aii py Die fists Ss el Va Ga 
 —Y =NnN4+ Pe hie rapes reece) 
 
 Consequently, if D denote the length of the perpendicular 
 intercepted between the point and the given straight line, 
 
 D = (Z - 2%)? +(X -wv)’?+(V-y’)’, 
 
30 
 
 ie CA ae 
 lt m+n 1+m+n 
 S24 m (a —a)+n(y'— b)}? 
 Hy 1+m+n 
 
 +? 4 (a -— a)? + (y' - 5)’, 
 
 ae re) ae) 
 
 39. The perpendicular distance of a point from a straight 
 line may also be expressed in terms of the angles which the 
 line makes with the co-ordinate axes. 
 
 Let AB (fig. 10) be the given line passing through a 
 point 4 whose co-ordinates are a, b, c, and making angles 
 a, 3, y with the axes of a, y, and x; also let AP make with 
 the same axes, angles a’, 9’, y’, P being the point with co- 
 ordinates a’, y’, x’, from which the perpendicular PB is to be 
 drawn. ‘Then 
 cos PAB = cosa cos a’ + cos 3 cos 3’ + cosy cosy’ 
 
 _ (#—a)cosa+ (y'—b) cosB+-(z" 
 W AP 
 . BP? = AP* — (AP cos PAB)? 
 = (wv — a)? + (y'— b)’ + (x — e)? 
 — {(w -— a) cosa + (y' = 6) cos B + (2’— c) cosy}* 
 
 Or, if we denote the cosines as usual by /, m, n, the distance 
 of the point w’ y’ x’ from the line 
 
 neyeoay SC Coe 
 
 ©-a y-b w-e 
 
 — —_ 
 —_—_—____ 
 
 is »/[(#—a)’+ (y’—b)? + (x'—c)?— {1(w'—a) + m(y'—b) +n(x’—c) }?]. 
 
 If the line passes through the origin, this becomes 
 
 J w+ y? +8? — (la’ + my’ + nz’). 
 
 40. As two straight lines in space, although not parallel, 
 may never intersect, it is a Problem which sometimes arises 
 to determine the shortest distance between them; it may be 
 solved by means of the following proposition. 
 
 The shortest line which can join any two points in two 
 straight lines in space, is the line which meets both of them 
 at right angles. 
 
31 
 
 Take one of the lines OC (fig. 11) for the axis of #, and 
 let the other 4B meet the planes of sw, vy, in A and B, in 
 which planes draw AO, Oy perpendicular to OC, and let 
 these be the axes of = and y. Draw By parallel to Ow, join 
 Ay and draw OM perpendicular to it; draw MN parallel to 
 By, make Ow equal to MN and join Nw. Then MN being 
 parallel to Ow is perpendicular to MO; therefore OM is 
 perpendicular to each of the lines WN, Ay, and therefore to 
 the plane 4yB; also MN being equal and parallel to Oa, 
 the figure VO is a rectangle, and Nw is equal and parallel to 
 MO; therefore Nw is perpendicular to Ow, and also to the 
 plane AyB and therefore to 4B; hence Naw meets each of 
 the given lines at right angles. It is also their shortest 
 distance; for take any point K in AB, draw KL parallel to 
 By and join OL; then the distance of K from O# is equal to 
 OL, because a perpendicular from K on Ow would be parallel 
 and equal to OL, zLOw being a right angle, and OL is 
 manifestly greater than OM. 
 
 41, Any plane passing through the shortest distance will 
 be perpendicular to the plane to which both the given lines 
 are parallel; for since Na is perpendicular to the plane Ay B, 
 any plane through Nw is perpendicular to dyB; and AyB, 
 since it contains one and is parallel to the other, is itself 
 parallel to every plane to which both the lines are parallel. 
 Also if through OC we conceive a plane to be drawn parallel 
 to AyB, it will be parallel to 4B; and MO = Ne will be its 
 perpendicular distance from the plane 4yB; i. e. the shortest 
 distance between two lines not in the same plane is the per- 
 pendicular distance of two planes, each of which is drawn 
 through one of the lines parallel to the other. 
 
 The complete determination of the shortest distance be- 
 tween two lines, by finding the position and magnitude of the 
 shortest distance when the equations to the lines are given, is 
 effected in Probs. 4 and 5. (Appendix, Section I.) 
 
SECTION II. 
 
 ON SURFACES OF THE SECOND ORDER. 
 
 42. Tue foregoing section comprehends all the more 
 useful results relative to lines and planes referred to rect- 
 angular co-ordinates. We shall now proceed to investigate 
 the equations, and forms, of surfaces of the second order, 
 which, after the plane, are next in simplicity; reserving for 
 the subject of the following section (after the reader has 
 become familiar with the mode of representing surfaces by 
 equations, and thence deducing their forms) one of the prin- 
 cipal means of modifying and simplifying those equations, 
 viz. transformation of co-ordinates. We shall next have oc- 
 casion to introduce the projections of lines and plane sur- 
 faces; and the more useful results, relative to the line and 
 plane referred to oblique co-ordinates, may be then more 
 conveniently given than at present. It will be necessary 
 however to make some preliminary observations on the geo- 
 metrical meaning of equations containing only one or two of 
 the variables wv, y, x, when we embrace in our enquiries the 
 three dimensions of space. 
 
 43. It was remarked (Art. 10) that the equation 
 =a, when three dimensions of space are regarded, repre- 
 sents an indefinite plane parallel to yz; similarly, the equa- 
 tion y? + py+q=0, which gives for y two constant values 
 y=a+b, represents two planes parallel to sw; and in 
 general, every equation containing only one of the variables, 
 represents a system of planes parallel to the two axes whose 
 co-ordinates it does not involve. 
 
 Again, the equation f(#, y)=0, which, in Plane Geo- 
 metry, belongs to a curve’ CC" (fig. 12) in the plane of vy, 
 in Geometry of Three Dimensions, represents a cylindrical 
 surface formed by drawing lines through all the points of CC’ 
 
 ae 
 
 ae ee 
 
33 
 
 parallel to Ox; for any point P in this surface, whatever be 
 the value of PC or x, will have the same 2 and y as its 
 projection C; and therefore the co-ordinates of every point 
 in the cylindrical surface will satisfy the relation f(a, y) = 0, 
 which does not contain x; and any point whose projection 
 is not in CC’ cannot by its co-ordinates satisfy the relation 
 Sf (@, y) = 0. The same observations apply to equations of 
 the form f(a, x) =0, f(y, x) =0. Hence, an equation con- 
 taining only two of the variables represents a cylindrical 
 surface parallel to that axis whose co-ordinate it does not 
 involve; and of which the trace upon the plane containing 
 the other two axes, is in Plane Geometry given by the same 
 equation. 
 
 44. The locus of an equation f(a, y, z) = 0, containing 
 all the variables, is a surface; that is, if all the points be 
 taken whose co-ordinates satisfy it, they will not comprehend 
 all the points of a solid figure, but only those situated in a 
 surface. In the equation f(x, y, s) = 0 we may assume any 
 values for two of the co-ordinates; and, deducing the value 
 of the third from it, we have the co-ordinates of the cor- 
 responding point in the locus; and similarly, we may de- 
 termine as many points in the locus as we please. Instead 
 of this, however, let us assume the value of only one co- 
 ordinate w=a; then (Art. 43) f(y, x, a) =0, represents a 
 cylindrical surface parallel to the axis of w; but since we 
 must only take the points in this surface which satisfy the 
 condition vw =a, it follows that by this assumption we obtain 
 a curve, namely, the section of the cylinder made by the 
 plane =a parallel to yx. Let this be PQ (fig. 13), 
 where ON =a; then of all the points in the indefinite plane 
 PNQ it is only those in the curve PQ which are deter- 
 mined by the equation f(y, x, a) =0; similarly, if we take 
 «= ON'=a, we shall have f(y, x, a’) = 0, which equations 
 determine the curve P’Q’; and, proceeding in this manner, 
 we may obtain an infinite number of curves all situated in 
 planes parallel to yx, and succeeding one another at intervals 
 as small as we please, the assemblage of which will form a 
 surface which is the locus of the equation f (7, y, x) = 0. 
 
 3 
 
34 
 
 45. We conclude, therefore, that every equation, whether 
 containing one, two, or three of the variables, represents a. 
 surface; if however the equation can be satisfied by no 
 system of real values of the co-ordinates, the surface will 
 be altogether imaginary; or, if it can be satisfied only by 
 breaking it up into two or three others, the surface is reduced 
 to a limited number of lines or points. In other cases the 
 equation will enable us to determine the figure and properties 
 of the surface which is its locus, as will be seen. 
 
 In discovering the figure from the equation, the traces 
 on the co-ordinate planes, determined by putting #, y, # 
 separately = 0 in the equation to the surface, and the sections 
 made by planes parallel to the co-ordinate planes, determined 
 by putting the co-ordinates each separately equal to a constant, 
 are the principal means. Thus (fig. 13) 4B, BC, CA are the 
 three traces of the surface; and PQ is a section made parallel 
 to yx, the equation to which is f(y, x, a) = 0, where ON = a, 
 and QN, NP are to be regarded as the axes to which the 
 curve PQ is referred. 
 
 As in Geometry of ‘wo Dimensions a curve may have 
 several branches, and an ordinate may have several values 
 corresponding to the same value of the abscissa; so a surface 
 may have several sheets, and points in different sheets may 
 have the same projection on the co-ordinate planes. As many 
 real values as x has for given values of w and y, so many 
 sheets will the surface have; if the value of x is imaginary, 
 there 1s no point in the surface corresponding to those values 
 of wv and y. 
 
 46. Surfaces, in the same manner as lines, are divided 
 into orders, according to the degree of their equations; the 
 degree being determined by the sum of the indices of the 
 three variables in that term of the equation (supposed to 
 contain no fractional or irrational term) where it is greatest. 
 The plane is the surface of the first order, being the locus 
 of the equation of the first degree between three variables ; 
 the sphere, the common cone and cylinder, the ellipsoid, &c., 
 are surfaces of the second order, because their equations are 
 
35 
 
 of the second degree. We shall at present obtain the equation 
 to each variety of the surfaces of the second order from some 
 known properties of them, and determine their figures; re- 
 serving the discussion of the general equation of the second 
 order, and the closer consideration of the properties of the 
 surfaces which it represents, to a more advanced part of the 
 work. ‘This arrangement, which corresponds to that usually 
 followed with regard to curves of the second order, will 
 gradually introduce the student to the more difficult parts of 
 the subject, and put him at once in possession of the more 
 important results. 
 
 47. 'To find the equation to the surface of a sphere. 
 
 The characteristic property of this surface is, that every 
 point in it is at the same distance from the center. 
 
 Let #, y; x be the co-ordinates of any point in the sur- 
 face, a, 6, c the co-ordinates of its center, and 7 its radius; 
 then 
 
 (vw -— a)? + (y— by +(e - cP? =7”", 
 
 is the relation which the co-ordinates of every point must 
 satisfy, and therefore the equation to the surface. 
 
 Cor. This equation will assume different forms accord- 
 ing to the position of the origin; thus if the origin be in 
 the center, a=b=c=0, and fie equation becomes 
 
 Pty tear; 
 if the origin be in the surface of the sphere, 
 a+bP+e=r", 
 and the equation becomes 
 v+y +s? =2 (av + by +c). 
 48. To find the equation to the surface of an oblique 
 cylinder, the base being circular. 
 
 This surface is generated by an indefinite line which is 
 carried round the perimeter of a given circle, always remain- 
 ing parallel to a given straight line, 
 
 3—2 
 
36 
 
 Let r=mz, y=ne, be the equations to a line OR 
 (fig. 14) through the origin, to which the generating line 
 is always parallel; and r the radius of the circle OPA 
 described in the plane of wy with its diameter coinciding 
 with the axis of x, along which the generating line moves. 
 Let PQ be the generating line in any position, and wv, y, 
 the co-ordinates of any point M in it, and therefore in the 
 cylindrical surface. Then the equations to PQ are 
 
 X-wv=m(Z—2), Y-y=n(Z—-s), (Cor. 2, Art. 21); 
 therefore, making Z = 0, we have the co-ordinates of Pjror 
 X =ON=2-mz, Y=NP=y- nz. 
 
 But, P being a point in the circular base, 
 Y?=2r X'— X”"; 
 therefore, substituting for Y’ and X” their values, 
 (y — nx)? =2r(# — mz) -—(# -— mz)’, 
 
 a relation among the co-ordinates of any point in the surface, 
 and therefore its equation. 
 
 Cor. We may suppose the curve OP, instead of a circle, 
 to be any curve of the second order determined by the equa- 
 tion Y°= 1X’ — (1 — e?) X”; we shall then obtain the equation 
 to any cylindrical surface of the second order, viz. 
 
 (y —nz)?=1(e — mz) —- (1 -— e&) (@—- mz)’. 
 
 And it is easily seen that whatever the curve OP be, if its 
 equation be Y’ = f(.X’), the equation to the cylindrical surface 
 generated by a straight line carried along it, and of which 
 it is called the directrix, will be (y — nz) = f(#@ — mz). 
 
 49. To find the equation to the surface of an oblique 
 cone, the base being circular. 
 
 This surface is generated by an indefinite line which is 
 carried round the perimeter of a given circle, always passing 
 through a fixed point. 
 
37 
 
 Let a, b,c be the co-ordinates OR, RQ, QV, (fig. 15) of 
 the vertex V, and r the radius of the circle OPA described 
 in the plane of wy with its diameter coinciding with the axis 
 of xv, along which the generating line moves. Let PV be the 
 generating line in any position, and a, y, x the co-ordinates of 
 any point M in it, and therefore in the conical surface ; then 
 the equations to VP are (Art. 21) 
 
 =f; 
 2.33) We ea rm ay 7A Sc) 
 z—C€ 
 
 xv 
 X-a= 
 x 
 
 therefore making Z = 0, we have the co-ordinates of P, or 
 
 z Sf 
 PON ae ere VON oe ee 
 
 wot C bs a 
 
 But, P being a point in the circular base, 
 Y= or XX"; 
 therefore, substituting and reducing, 
 (bz —cy)’ = 2r (x — c) (ax — cx) — (ax —c2)?, 
 the equation to the conical surface. 
 
 Cor. We may suppose the curve OP, instead of a 
 circle, to be any curve of the second order, determined by 
 the equation Y? = 1_X’— (1 — e) X”; we shall then obtain the 
 equation to any conical surface of the second order, viz. 
 
 (bz —cy)’=1(% —c) (az -— ca) — (1 - &) (az - € 2)’. 
 
 And it is manifest that whatever the curve OP be, if its 
 equation be Y’=f(X’), the equation to the conical surface 
 of which it is the directrix will be 
 
 bz—cy aS— Cx 
 ra alereces 
 50. To find the equation to the surface of a spheroid. 
 
 This surface is generated by the revolution of an ellipse 
 about one of its axes. 
 
38 
 
 Let CA (fig. 16) be a quadrant of an ellipse which, by 
 revolving about its axis OC coinciding with the axis of x, 
 generates a spheroid. Let A’C be any position of the gene- 
 rating curve, and ON =7, NM =y, MP’==3z the co-ordinates 
 of any point P’ in it; AO=a, CO=c; then from the right- 
 angled triangle ONM, 
 
 OM = a aa y’. 
 But, since P’ is a point in an ellipse with 3 axes a and ec, 
 
 OM? 3 
 weg e om opt. 
 o 
 
 the equation to the surface; which is called an oblate or 
 prolate spheroid, according as the axis 2c, about which the 
 ellipse revolves, is the less or greater of its axes. 
 
 Cor. Similarly, if the hyperbola 4Q (fig. 18) revolve 
 about its conjugate axis OC coinciding with the axis of gz, 
 the equation to the surface of the hyperboloid generated, 
 consisting of one continuous sheet, will be 
 
 2 2 
 very 
 Pe: 
 
 and if it revolve about its tranverse axis OA (fig. 19) co- 
 inciding with the axis of aw, the equation to the surface 
 generated, consisting of two disunited portions or sheets, 
 will be 
 | yt 
 
 DB) ° 
 
 a~ Ce 
 
 Ya 
 
 51. To find the equation to the surface of a paraboloid. 
 
 Let OQ (fig. 20) be a parabola which by revolving about 
 its axis, coincident with the axis of w, generates a paraboloid ; 
 OP any position of the generating curve, 
 
 ON =2, NV HW, eM Lee, 
 
39 
 
 the co-ordinates of any point P in it; then 
 PN’ =1.0N = la, 
 1 being the latus rectum of OP; but from the right-angled 
 triangle PMN, 
 PN? me y” a 2", 
 YY te = le, 
 
 the equation to the surface. 
 
 52. When, as in the preceding instances, a plane curve 
 revolves about an axis coincident with one of the co-ordinate 
 axes, the equation to the surface generated may be readily 
 obtained, whatever be the generating curve. For let OC 
 (fig. 16) the axis of x be the axis of revolution, CPA the 
 generating curve, and PR =f(OR) its equation; and let it 
 revolve about OC into the position CP’4’; ON =a, NM = y, 
 MP’ = x, the co-ordinates of any point P’. 
 
 Then by the equation to the curve, MP’ =f(OM), 
 or x= f(a" + 9"), 
 
 the equation to the surface. 
 
 53. The surfaces which we have hitherto considered, 
 are the simplest cases of surfaces of the second order; their 
 equations are all contained in the general equation of the 
 second order between three variables, which is 
 
 av +by? +e 42a yr+20 sx4+2cuyt2a v4+2b y+20"s+4+d=0. 
 
 Tt will be shewn hereafter that this equation, by giving 
 a proper position and directions to the origin and axes of 
 the co-ordinates, can always be reduced to one of the forms 
 
 Av? + By + C2? =+ D, 
 By + C2? =2A'e, 
 
 which represent two distinct families of surfaces; the former 
 
40 
 
 those which have a center, the latter those which have not 
 a center; meaning by center, a point such that all chords 
 of the surface drawn through it are bisected in it. 
 
 54. The origin is the center, and the co-ordinate planes 
 are the principal planes, of the surface represented by the 
 equation 
 
 Aa’ + By? + Cz’ = + D. 
 
 For let P (fig. 22) be a point in the surface whose 
 co-ordinates are v=h,. y=k, x =1; then the equation is 
 satisfied by these values, and therefore it is also satisfied 
 when —h, —k, —/ are written for w, y, and x. But if we 
 produce PO to P’, and make OP’= OP, the co-ordinates of P’ 
 are evidently equal to —h, —k, —1; therefore P’ is a point 
 in the surface, and PP is a chord, and it is bisected in O; 
 i.e. every chord is bisected in O, therefore O is the center of 
 the surface. 
 
 Also the surface is situated symmetrically with respect 
 to the co-ordinate planes. For if in the equation we make 
 v«=ON=h, and y= NM=k, it will give for = two equal 
 values with contrary signs PM, QM; so that for every point 
 of the surface, situated above the plane of wy, there will be 
 a corresponding point situated at an equal distance below it. 
 Similarly, the other co-ordinate planes may be shewn to bisect 
 all their ordinates at right angles. Planes which have this 
 property are called the principal planes of the surface, and 
 their intersections (which in this case coincide with the axes 
 of the co-ordinates, and with respect to each of which also 
 the surface is symmetrically situated) the aves of the surface. 
 Also every chord passing through the center is called a diameter 
 of the surface. 
 
 55. In the equation to surfaces that have a center 
 Aw’ + By’ + Cx2? = +D, 
 
 having taken care to make the second member positive, since 
 the coefficients of the variables cannot be all negative together, 
 
41 
 
 we can only have three varieties of form, (1) all the coefficients 
 positive, (2) one negative, (3) two negative; so that the equa- 
 tion may assume the three following forms : 
 
 2 2 2 
 
 Lv yo 
 
 nee + F + re ! 9 
 oO 5 
 
 ian Lene 
 
 x’ y 2 
 
 ates POR a 
 
 pa: y? oe? 
 
 The surfaces represented by them are called respectively the 
 ellipsoid, the hyperboloid of one sheet, and the hyperboloid 
 of two sheets; the two latter surfaces being-also sometimes 
 called the continuous, and discontinuous hyperboloid. 
 
 56. In the family of surfaces, represented by the equation 
 By’? + Cz? = 242, 
 
 the origin is not the center, since the equation does not remain 
 unchanged when w, y, x are changed into —w, —y, —#; and 
 it will be seen hereafter that no other point can be its center. 
 Also since only even powers of y and x enter the equation, 
 the planes of sw and wy are principal planes, but the plane 
 of yx not a principal plane; consequently the surface has 
 only one axis, coinciding with the axis of vw. In the equation, 
 if Band J’ are not both positive, let the coefficient of y? be 
 made positive; and then, if necessary, change the sign of & 
 in order to make the second member positive, for the change 
 of vw into —@ will merely alter the position of the surface ; 
 the equation will then offer two varieties according as C is 
 positive or negative, under the forms 
 
 2 2? 
 Tipe 
 y” 2 
 
 1 ie v 3 
 
 and the surfaces represented by them are called respectively 
 the elliptic and hyperbolic paraboloids. 
 
42 
 
 57. To find the equation to the surface of an ellipsoid. 
 
 This surface is generated by a variable ellipse which 
 moves parallel to itself with its axes in two fixed planes, 
 and vertices in two ellipses in those planes having a common 
 axis coincident with the intersection of the planes. 
 
 Let BC, CA (fig. 17) be quadrants of the given ellipses 
 traced in the planes yx, sv”; OC =e their common 3 axis 
 coinciding with the axis of zy, OA =a, OB =b, the other 
 4 axes; QPR a quadrant of the generating ellipse in any 
 position, having its plane parallel to wy, its center in OC, 
 and two of its vertices in the ellipses AC, BC, so that the 
 ordinates QN, RN are its 4 axes; also let ON=x, NM =a, 
 MP =y be the co-ordinates of any point P in it; then 
 
 x” y 
 
 2 
 
 in ere ae 1 
 
 NQ? NR () 
 NQ? zs NR Pte 
 
 but AG es ——=1--—. 
 a’ Ce : Cc 
 
 TL)? 
 
 Multiply the first term of (1) by 
 
 » the second term by 
 
 ae and the second member of the equation by the equal 
 
 Pei 
 quantity 1 aKa and transpose and we get 
 
 2 2 2 
 (ite ATs ie 
 =+3+- 
 
 =}, 
 ‘Tie 8 SSO pe 
 
 the equation to the surface. 
 
 58. To determine the form of the ellipsoid from its 
 equation. 
 
 Since in the equation, w can only receive values between 
 a and —a, y between 6b and — 6, and z between ec and —e, 
 the surface is limited in all directions. If we put x = 0, 
 we obtain 
 
43 
 
 for the equation to the trace on wy, which is therefore an 
 ellipse AB; also, from the mode of generation, all sections 
 parallel to vy are ellipses, and since their axes are in the 
 ratio of a to b, they are all similar to AR. 
 
 If we make x = +h, we have 
 
 aphTe Bigs® h? 
 
 ~=+—sl-- 
 
 bh? Ce a’ 
 for the equation to any section parallel to yz, which is an 
 ellipse similar to the trace BC (since its axes are in the 
 ratio of b to c whatever be the value of h) and which becomes 
 imaginary when h>a. In the same manner it may be shewn, 
 that all sections parallel to xsw-are ellipses, similar to the 
 trace AC. The ellipses, of which 4B, BC, CA are quadrants, 
 in which the surface is intersected by the co-ordinate planes, 
 which are its principal planes, are called the principal sections 
 of the surface; and the parts of the co-ordinate axes inter- 
 cepted within the surface, are called its axes. Hence a, b, c 
 represent the J axes of the ellipsoid, and also the 4 axes of 
 its principal sections. The extremities of the axes such as 
 A, B, C are called the vertices of the ellipsoid, of these it has 
 six, one at the extremities of each axis. 
 
 The whole surface consists of eight portions precisely 
 similar and equal to that represented in the figure. 
 
 Cor, If a=b, the equation becomes that to a spheroid 
 generated by revolution about Ox; similarly, if any other two 
 of the semiaxes become equal, the ellipsoid becomes a spheroid 
 generated by revolution about the remaining axis. 
 
 59. To find the equation to the surface of a hyper- 
 boloid of one sheet. 
 
 This surface is generated by a variable ellipse which 
 moves parallel to itself, with its axes in two fixed planes, 
 and vertices on two hyperbolas in those planes having a 
 common conjugate axis coincident with the intersecticn of 
 the planes. 
 
44. 
 
 Let AQ, BR (fig. 18) be the given hyperbolas traced 
 in the planes xa#, yx; OC =c their common 4 conjugate 
 axis coinciding with the axis of xs, OA4=a, OB=b, the 
 4 transverse axes; QPR the generating ellipse in any posi- 
 tion, having its plane parallel to wy, its center in OC, 
 and its vertices in the hyperbolas 4Q, BR, so that thie 
 ordinates NQ, NR, are its 4 axes. Also let ON = 2, 
 NM=.2a, MP =y be the co-ordinates of any point P in 
 
 the generating ellipse; then 
 
 x y” 
 NO}? NR (1) 
 NQ? SONAR ed 
 but =i eas Sie it ye 
 
 2 
 
 Multiply the first term of (1) by zu , the second term by 
 
 a” 
 2 
 
 , and the second member of equation (1) by the equal 
 
 b? 
 2 
 quantity as and transpose and we get 
 
 a y” 
 a iit pestered 
 
 the equation to the surface. 
 
 60. To determine the form of the hyperboloid of one 
 sheet from its equation. 
 
 Since the equation will visibly admit values of a, y, z 
 positive and negative however large, the surface is extended 
 indefinitely on all sides of the origin, If we put #=0, 
 
 we obtain 
 2 
 
 | & 
 
 y 
 4 
 
 +21 
 
 © 
 w 
 
 g 
 
 for the equation to the trace on wy, which is the ellipse 4B; 
 and from the mode of generation all sections parallel to wy are 
 similar ellipses, the dimensions of which increase indefinitely, 
 
45 
 
 the least being that of which AB is a quadrant, and which 
 forms the interior limit of the surface. For the sections 
 parallel to yx, putting # = +h, we have 
 
 which, as long as h<a, and therefore the second member 
 positive, represents a hyperbola similar to the trace BR, with 
 its vertices in the ellipse AB, and conjugate axis parallel to 
 Ox, since the coefficient of x’ is negative. When h=a, the 
 equation represents two straight lines; and when h >a, 
 making the second member positive, the equation is 
 
 y’ h? 
 
 ee at 
 which also represents a hyperbola, but in a new position, 
 namely, with its vertices in AQ, and conjugate axis parallel 
 to Oy. In the same manner, the sections parallel to zx” may 
 be shewn to be hyperbolas similar to the trace 4Q. The 
 principal sections of this surface are the ellipse 4B, and the 
 
 hyperbolas 4Q, BR, 
 
 The quantities a, b, which denote the distances from the 
 origin at which the surface cuts the axes-of w and y, are 
 called the real semiaxes of the surface; the quantity ¢ is 
 called the imaginary semiaxis, because if we put 2 = 0, y = 0, 
 to find where the surface cuts the axis of x, we find 
 
 v= —c, or s=+0e\/-1, 
 
 so that c is the coefficient of the expression for the imaginary 
 4 axis of the surface. ‘The extremities of the real axes are 
 called the vertices of the surface; two of them are 4, B, and 
 the remaining two are at distances from O respectively equal 
 to a and 6, in AO and BO produced. The whole surface 
 consists of eight portions, precisely similar and equal to that 
 represented in the figure; and since it is continuous, that is, 
 since we can pass from one point in it to any other point in it, 
 without quitting the surface, it is called the hyperboloid of 
 one sheet. 
 
46 
 
 Cor. If a= 8, all sections parallel to wy become circles, 
 and the surface becomes a hyperboloid of revolution about the 
 conjugate axis. 
 
 61. The hyperboloid of one sheet has an interior conical 
 asymptote. 
 
 Putting the equation under the form 
 
 3 x? y° x yy a’*b? 
 char ae ee pas) La 5 ap a 
 PA hace b- ad 6 ay” + bx’ 
 
 and expanding the value of ‘a by the Binomial Theorem, we 
 c 
 
 see, since when a and y are very great the quantity 
 a’ b? 
 a*y* of b? a? 
 
 is very small, that the relation among the co-ordinates of 
 points very distant from the origin, is nearly expressed by 
 
 a” 2 
 
 Poteet ds 
 
 CG amos 
 
 this then is the equation to a surface, whose distance from the 
 surface of the hyperboloid, measured parallel to the axis of z, 
 diminishes indefinitely as w and y increase; and, as its 
 ordinate is always greater than the corresponding ordinate of 
 the hyperboloid, it lies within the latter surface. 
 
 This asymptotic surface is a right cone vertex O, and base 
 an ellipse A’B’, whose center is in C, and 3 axes equal and 
 parallel to 40, BO; for if x, y, be co-ordinates of any point 
 in a generating line of this conical surface, its equations will 
 
 be 
 Ye a7 ae Ze 
 & be 
 
 . the co-ordinates of the point where it meets 4’B’ are, 
 making Z=c, 
 
47 
 
 and as these are co-ordinates of a point in an ellipse whose 
 4 axes are a and 6, 
 
 “ve\? yc” Pa Yad 2 
 Za 3b CF 5 ae be 
 
 62. To find the equation to the surface of a hyperboloid 
 of two sheets. 
 
 This surface is generated by a variable ellipse which 
 moves parallel to itself, with its axes in two fixed planes, 
 and vertices in two hyperbolas in those planes having a 
 common transverse axis, coincident with the intersection of 
 the planes. 
 
 Let AQ, AR (fig. 19) be the given hyperbolas traced in 
 the planes xv, wy; OA =a, their common 4 transverse axis 
 coinciding with the axis of 2, OB=b, OC =e, the 4 con- 
 jugate axes; QPR the generating ellipse in any position, 
 having its plane parallel to yz, its center in Ow, and its 
 vertices in AQ, AR, so that the ordinates QN, RN are its 
 semiaxes. Let ON=a2, NM=y, MP==2 be the co-ordi- 
 
 nates of any point P in the ellipse, then 
 
 3° y° 
 Qn? + RNeT ? M) 
 pay ahs RN* x 
 but Q —— Out 9 9 aay 2 1 2 
 a b- a 
 
 2 
 
 multiply the first term of equation (1) by 
 
 —, the second 
 Cc 
 
 2 
 
 RN° 
 term by aprrat and the second member of equation (1) by 
 
 o 
 
 - 
 
 ES ao 
 the equal quantity — — 1, and transpose and we get 
 me 
 
 the equation to the surface. 
 
48 
 
 63. To determine the form of the hyperboloid of two 
 sheets from its equation. 
 
 The equation shews that all values of v, between + a and 
 —a, are inadmissible, therefore no part of the surface can be 
 situated between two planes parallel to yx through A and A’; 
 but the equation can be satisfied by values of wv, y, % in- 
 definitely great, therefore there is no limit to the distance to 
 which the surface may extend from O. If we put # =0, we 
 have 
 
 2 
 
 Dae ; 
 
 therefore the principal section by yz is imaginary; but all 
 sections parallel to yz, and at a distance from it greater 
 than a, are similar ellipses as appears from the mode of 
 generation. For the sections parallel to wy, putting x= +/, 
 
 we have 
 2 I? 
 
 — 
 
 eb Te? 
 which represents a hyperbola similar to the principal section 
 AR with its vertices in AQ and the opposite branch, and 
 conjugate axis parallel to Oy; and in the same way it may 
 be shewn, that the sections parallel to xv are hyperbolas 
 similar to the principal section AQ with vertices in AR and 
 the opposite branch, and conjugate axes parallel to Ox. 
 
 If we make at once y = 0, x = 0, to find where the surface 
 cuts the axis of w, we have 2 = a? or a = £a3 2a is the real 
 axis of the surface, and its extremities the vertices of the 
 surface of which there are but two; the quantities 6b, ¢ are 
 the imaginary semiaxes of the surface, because it does not 
 cut either the axis of y or x. ‘The whole surface consists 
 of two sheets perfectly similar and equal, and indefinitely 
 extended in opposite directions, but separated by an interval 
 in which exists no point of the surface ; it is therefore called 
 the hyperboloid of two sheets; each sheet consists of four 
 portions precisely similar and equal to that represented in the 
 
 figure. 
 
49 
 
 Cor. If 6 =c, or the two imaginary axes become equal, 
 this surface becomes a hyperboloid of revolution about the 
 transverse axis. 
 
 64. The hyperboloid of two sheets has an exterior 
 conical asymptote. 
 
 Putting the equation under the form 
 
 x y° sig 3° : (4 =) ( b? Cc ) 
 = 0 as tree Bad ede reer Witeustna y Can 
 i ON bans Ce ey? + b?2*)” 
 
 and expanding the value of — by the Binomial Theorem, we 
 a 
 
 have, for points very distant from the origin, the relation 
 among the co-ordinates nearly expressed by 
 
 this then is the equation to a surface whose distance from the 
 hyperboloid, measured parallel to the axis of w#, diminishes 
 indefinitely as y and s increase; and, as its. ordinate w is — 
 always less than the corresponding ordinate of the hyperbo- 
 loid, the latter surface lies within it. It may be shewn, as 
 in Art. 61, that the asymptotic surface is a right cone vertex 
 O, and base an ellipse B’C’ whose center is A and semi-axes 
 parallel and equal to OB, OC. In this case, as in the former, 
 we observe that the equation to the conical asymptote is ob- 
 tained by omitting the constant term in the equation to the 
 surface. 
 
 65. Since from the equation to the ellipsoid, the equa- 
 tion to the hyperboloid. of one sheet results by changing ce 
 into c\’—1; and the equation to the hyperboloid of two 
 sheets, by changing 0b into b,\/—-1 and ¢ into CAL sevais if 
 any result in terms of its axes be obtained for the ellipsoid, 
 the corresponding results for the hyperboloids may be de- 
 duced by writing 
 
 GA for c; or b /—1, ae at, for 6 and c 
 4 
 
50 
 
 66. To find the equation to the surface of the elliptic 
 paraboloid. 
 
 This surface is generated by a parabola which moves with 
 its plane perpendicular to a fixed plane, and axis in that plane, 
 and parallel to the axis of another parabola along which its 
 vertex moves; the concavities of the parabolas being turned 
 towards the same parts. 
 
 Let OR (fig. 20) be a parabola traced in the plane of 
 wy, vertex at the origin, and axis coinciding with the axis 
 of «, J its. Jatus rectum; RP the generating parabola in any 
 position with its plane parallel to zw, vertex in OR, and axis 
 parallel to Ow, and let /’ denote its latus rectum, and ON = a, 
 
 JM=y, MP =z be the co-ordinates of any point P in it; 
 also draw RM’ parallel to Oy. Then 
 
 2 
 
 s=l'.RM =U (ON - OM’) =! (« _ =| 
 
 9 
 
 Neen aye 
 f 7a 
 
 the equation to the surface. 
 
 67. To determine the form of the elliptic paraboloid 
 from its equation. 
 
 Since only positive values of w are admissible, no part of 
 the surface is situated to the left of the plane of yx; also 
 since the equation can be satisfied by positive values of w, and 
 by positive and negative values of y and x however large, the 
 surface is extended indefinitely towards the positive direction 
 of w If we make y=0, 3° =/w is the equation to the 
 principal section OQ, and from the mode of generation all 
 sections parallel to xv are parabolas equal to OQ with vertices 
 in OR; similarly, all sections parallel to wy are parabolas 
 equal to the other principal section OR, and having their 
 vertices in OQ. If we make w =0, we find I'y? + /z*=0; 
 *. y=0, x=0, or the trace on yg is a point namely the origin. 
 If we make w =h, we have 
 
 y” 3? 
 
 TM 
 
51 
 
 for the equation to a section parallel to yx, which represents 
 an ellipse whose semiaxes are the ordinates QN, NT' which 
 
 are to one another in a constant ratio \//' to \/1; therefore 
 all sections perpendicular to the axis of the surface, are similar 
 ellipses, and hence the name of the surface. It has only one 
 vertex O and one indefinite axis Oa, and consists of one sheet ; 
 and is reduced to a paraboloid of revolution when J =U’. 
 
 68. To find the equation to the surface of the hyperbolic 
 paraboloid. 
 
 This surface is generated by a parabola which moves with 
 its plane perpendicular to a fixed plane, and axis in that plane, 
 -and parallel to the axis of another parabola along which its 
 vertex moves; the concavities of the parabolas being turned 
 towards opposite parts. 
 
 Let OR (fig. 21) be a parabola in the plane of wy, vertex 
 at the origin, and axis coinciding with the axis of w; and J its 
 latus rectum; HP the generating parabola in any position 
 with its plane parallel to x7, vertex in OR, and axis parallel 
 to Ox, and let /’ denote its latus rectum, and ON=a, NM=y, 
 MP = the co-ordinates of any point P in it; draw RM 
 parallel to Oy, then 
 
 o 
 
 ~ 
 
 2 =I.MR =! (OM -ON)=0 (F- w) 
 
 2 
 l 
 
 the equation to the surface. 
 
 69. To determine the form of the hyperbolic paraboloid 
 
 from its equation. 
 
 The surface cuts the co-ordinate axes only at the origin, 
 and since the equation will visibly admit values of aw, y, % 
 positive and negative however large, the surface is extended 
 indefinitely from the origin. If we make y=0, we have 
 s* = —I'x for the equation to the trace on xa, which repre- 
 
 4 —2 
 
52 
 
 sents the parabola OQ, with its concavity turned towards the 
 negative direction of a, since its equation shews that w must 
 be taken negatively; and from the mode of generation, all 
 sections parallel to s# are parabolas equal to OQ, with their 
 vertices in OR. The other principal section of the surface in 
 the plane of wy, is the parabola OR, forming the interior limit 
 of the figure, and all sections parallel to vy are parabolas equal 
 to OR, with their vertices in OQ. For the trace on ya, 
 putting wv = 0, we have EVA sabat AWA which represents two 
 straight lines passing through the origin; and for sections 
 parallel to yz, making x = h, we have 
 
 which represents a hyperbola with its vertices in OR, and 
 conjugate axis parallel to Oz. If we make h negative, the 
 equation to the section becomes 
 
 » 9 
 8" Ji 
 
 Uh th | 
 which also represents a hyperbola, but in a new position, 
 viz. with its vertices in OQ, and conjugate axis parallel to 
 Oy. The surface consists of one sheet, and has only one 
 vertex O, and one indefinite axis Ow; and all sections per- 
 pendicular to its axis are similar hyperbolas, whence its name. 
 It does not become a surface of revolution when 7 = 7’, nor in 
 
 any other case. 
 
 b) 
 
 79. The hyperbolic paraboloid has plane asymptotes. 
 
 Putting the equation under the form 
 \é , 
 U Lx 
 v=-y—lea =F (2 - =) ; 
 
 and expanding the value of x by the Binomial Theorem, we 
 have the relation, among the co-ordinates of points very 
 distant from the origin, nearly expressed by 
 
53 
 
 is the equation to two planes through the origin perpendicular 
 to the plane of yx, the distance between which and the para- 
 boloid continually diminishes. These planes contain the 
 asymptotes to all the hyperbolic sections parallel to yz, which, 
 as we have seen, have their centers in the axis of #, and axes 
 
 in the ratio of Vs to / le 
 
 71. The equation to the hyperbolic differs from that to 
 the elliptic paraboloid, in having — 7’ instead of J’; this re- 
 lation will enable us to modify all results obtained for one 
 surface, so as to be true for the other. 
 
 72. The elliptic and hyperbolic paraboloids are par- 
 ticular cases of the ellipsoid, and hyperboloid of one sheet 
 respectively ; viz. when the centers of these surfaces are 
 removed to an infinite distance. 
 
 In the equation 
 
 write vw —a instead of wx, then the resulting equation 
 
 2 
 
 (w-a)y ¥? ON ety 
 ceemeese e ee AT, gree oe fe O 
 
 a hous ct ei Ge i Ciden On Gi da 
 is reckoned from the vertex of the surface. Let p and p’ 
 denote the distances of the foci of the principal sections in 
 
 wy and xa, from the vertex, 
 . P=ad—(a-p)=2ap-p, e=2ap sp’; 
 
 hence, by substitution, 
 
 a” 2a y - 
 re Fe 9 , 7 = 0, 
 a a 2ap —p 2ap) += Pp 
 a y x 
 or ——-2@+ = 5 m= OS 
 MP 
 AY 1 lara Ay UN ae 
 a a 
 
 therefore, making @ infinite, 7,e. supposing the center of the 
 surface to remove to an infinite distance from the vertex, 
 
54 
 
 whilst the distances of the foci of the principal sections from 
 their common vertex remain finite, we have 
 2 2 
 Ee — ,—~2xH=0, 
 2p 2p 
 which coincides with the equations to the paraboloidal surfaces. 
 
 Cor. Hence if any result be obtained for the ellipsoid 
 or hyperboloid, it will be adapted to the paraboloids, by the 
 modification above indicated; viz. by transferring the origin 
 to the extremity of an axis, and making that axis infinite. 
 Also both families may be represented by the equation 
 
 Ax’ + By? + Cz? = 242, 
 
 the origin being at a vertex, and A = 0 when the surface has 
 not a center. 
 
 feectilinear generating lines of surfaces of the second order. 
 
 Surfaces of the second order admit of another division, 
 viz. into those which can be generated by the motion of a 
 straight line, and into those which cannot. This property, 
 which we have seen to belong to the cylinder and cone, we 
 shall now shew to be possessed by the hyperboloid of one 
 sheet and the hyperbolic paraboloid; it is easily foreseen that 
 it cannot belong to the remaining surfaces of the second order, 
 as their forms manifestly preclude their having a straight line 
 applied to them throughout its indefinite length; this will also 
 appear from our results. 
 
 73. The Hyperboloid of one sheet can have an infinite 
 number of straight lines entirely coinciding with its surface. 
 
 Since the equation to the surface may be written 
 
 2 2 
 v 2 y 
 a b 
 
 2 
 Ee) 
 
 w 
 
 it is evident that either of the following systems (each con- 
 sisting» of the equations to two planes and therefore represent- 
 ing a straight line) will satisfy it, viz. 
 
v& R 4 
 or ~ = Fay (1 +4) 
 @*- Cc b 
 
 w being an arbitrary constant; since then the co-ordinates of 
 every point in either of the lines (1) or (2) will satisfy the 
 equation to the surface by causing its two members to become 
 identical, it follows that for every value of mu there are two 
 straight lines that lie entirely in the surface. 
 
 74. ‘To determine the equations to the projections of the 
 generating lines of a hyperboloid of one sheet. 
 
 Let the equations to the surface and to a line be respectively 
 x” y 3 
 erg rhe oh 
 Ta eb 6 : 
 vw=ms+h, y=nsr+k; 
 if the line coincide with the surface in all its points, the 
 equation 
 (ms+h)? (nz+hkh) 
 GOLgey es Bo peer 3 
 a” b Cc 
 will be true for all values of x; therefore, equating to 
 nothing the coefficients of z* and z, we have 
 
 m oi 1 
 
 ae =O 
 
 a” ieee Ca : 
 
 mh nk ta Mi ; 
 riggs sae ee ie eet 
 ae b- > ef BP 
 
 Eliminate & between the two latter equations, and reduce 
 by means of the former ; 
 
 h? mb h\? RENS 
 2 aa 4 — +e = te or ( = 1% 
 a~ na a VUaAC 
 
 b 
 Dis ie BEG eat Me mhe 
 
 > 
 
56 
 
 therefore the two lines whose equations are respectively 
 
 nae NaC 
 C= MZ + — &=>M2Z — — 
 £ b 
 (1) += -Hh2)> 
 mbe mbe 
 y= NZ —- — Y = ns + — 
 a a 
 
 will coincide with the surface of the hyperboloid in all their 
 points, m and m being any quantities which satisfy the 
 equation 
 
 m 1 
 =r 5 
 Cc” 
 
 and if m be made to assume all values, and m be always 
 
 6 ; ae 
 taken = — \/a? = m2c?, we shall determine two infinite sys- 
 ac 
 
 tems of lines represented by the equations (1) and (2), 
 having the aforesaid property. 
 
 Cor. No two lines in the same system intersect; for 
 if we take two lines in the first system represented by the 
 equations 
 
 NAC fe NIE 
 L=MSs + — C= MS + 
 ? , 
 mbe f m be 
 y= ns — —— Y=NZ— 
 a J a 
 
 we shail find (Art. 22) that they cannot intersect unless 
 (m—-m’)?=0, that is, unless they become identical. But 
 any line in the first system intersects every line in the 
 second system; for let the equations to two lines, one in 
 each system, be 
 
 nac Welin ae 
 = MZ +—— L=EMZ— 
 b b 
 ? , 2 
 mbe ; mbe 
 Y= 2S = a Y= 22. -.—— 
 a 
 
57 
 
 24 fo ry; 
 ? 2 2 
 
 . ‘ : At Li: 
 then (Art. 22) these lines intersect if — call Ghia 
 a a 
 
 ay 1 
 which is always the case because each equals —. 
 c 
 
 Hence if we take any three lines in the second system, 
 and suppose them fixed, and make a line intersect them, it 
 must be a line of the first system, and by assuming all possible 
 positions will generate the surface of the hyperboloid; and 
 since consecutive positions of the generating line do not 
 intersect one another, the surface is what is called a twisted 
 surface. The surface evidently admits a second mode of 
 generation, in which a line belonging in every position to the 
 second system, moves along three fixed lines of the first 
 system. 
 
 75. The projections of the generating lines of the hyper- 
 boloid upon the principal planes, are tangents to the traces of 
 the surface on those planes. 
 
 Suppose w = mz + h to be the equation to a line touching 
 the principal section on sv whose equation is 
 
 9 2 
 
 ri fa & 
 
 then cial SAAN as We 
 
 must give two equal values for x, and 
 
 Pi ms +h)? 
 5 ae ME Ok: +1 must be a perfect square ; 
 
 1 mM? h? m? h? . b Peps 
 es Spaete ort FET = 4 , or h*=a° — mc’; 
 cb a* a a 
 
 e w=me+/ae—m ec, 
 
 which coincides with the equations to the projections of the 
 generating lines on zw. Similarly, the projections on the 
 
58 
 
 other principal planes may be shewn to be tangents to the 
 traces on those planes. 
 
 Cor. If through the origin we draw a line parallel to 
 the generating lines, its equations will be v=ms, y=nzx3 
 and the equation to the surface generated by it, eliminating 
 m and » by means of the equation which connects those 
 quantities, will be 
 
 which represents the conical asymptote; hence any line drawn 
 through the center parallel to a generating line of the hyper- 
 boloid lies in the surface of the conical asymptote. 
 
 76. The hyperbolic paraboloid can have an_ infinite 
 number of straight lines entirely coinciding with its surface. 
 
 It is evident that either of the following systems (each 
 consisting of the equations to two planes, and therefore repre- 
 senting a straight line,) will satisfy the equation to the 
 surface; viz. 
 
 oo. ¥ % y x 
 
 VLR tte an ae 
 
 Ore Ae UD Pike keer eae es (2) 
 
 DRT ARE 5 Ao A a 
 Since then the co-ordinates of every point in either of the 
 lines (1) and (2) will satisfy the equation to the surface by 
 causing its two members to become identical, it follows that 
 
 for every value of the arbitrary constant p, there are two 
 straight lines that lie entirely in the surface. 
 
 77. ‘To determine the equations to the projections of the 
 generating lines of a hyperbolic paraboloid. 
 
 Let the equations to the surface and to a straight line 
 be respectively 
 
59 
 
 e=met+h, y=nsr+k; 
 
 then, if the line coincide with the surface in all its points, 
 the equation 
 
 (nz+k)y 
 eo eae tae 
 
 ms+-h= 
 ti i j 
 is satisfied for all values of x; 
 are pa} i Qnk ; toe 
 =a TFT =U N= = <5 
 aoe Tieaee ttre 
 t 2 
 caper Ty fe py 
 2n 4, 
 
 will coincide with the surface of the paraboloid in all their 
 . 1 
 points, m being any quantity, and ” = NES and if m be 
 
 made to assume all values, we shall determine two infinite 
 systems of lines represented by equations (1) and (2), having 
 the aforesaid property. 
 
 Cor. In the same manner as for the hyperboloid, it 
 may be shewn that two lines in the same system never inter- 
 sect, and that two lines in different systems always intersect. 
 Hence if we suppose three lines in either system to become 
 fixed, the surface may be generated by making a line move 
 so as always to intersect them. Or, since the generating lines 
 in the two systems are respectively parallel to the fixed planes 
 yY=Nn2%, y= —nz, the paraboloid may be also generated in 
 two ways by a straight line which moves so as always to 
 Intersect two fixed lines, and to be parallel to a fixed plane. 
 
60 
 
 In this case also the projections of the generating lines are 
 tangents to the principal sections of the surface. 
 
 Thus (changing the sign of w in both equations) 
 —x=mz+h will bea tangent to x* =/'a, 
 
 if 2° + Imzx+Uh=0 bea perfect square, 
 ay! 
 or 40h =1?m’®, orh= aa 
 
 (+7) 
 . -—-& =m  - — 
 fe 
 
 is the equation to a tangent; which, measuring # in the 
 positive direction, coincides with the equation to the projection 
 of the generating lines on zw. 
 
 78. We shall terminate this Section with demonstrating 
 the following general and important property of surfaces of 
 the second order. 
 
 If two surfaces of the second order have a plane section 
 in common, their other curve of intersection, if it exist, will 
 
 also be a plane curve. 
 Let the equations to the two surfaces be 
 Aa’ + By? +C2?° 4+ 2A y2z4+2B' s24+2Cay+2A"v+2B’y 
 +20°2+ D=0, 
 au + by +cx? + 2a'yx 4+ 2b’ sax + Qcuxy 42a" v4 2b'y 
 + 2c"24d=0, 
 
 and suppose them to have a common section in the plane 
 of wy; then making x= 0, the curves represented by the 
 
 equations 
 Ax’? + By? +2Cay+2A"v+2B"y+D=0, 
 
 ax+by+2ceary 42a a+ 2b y+d=0, 
 
61 
 
 are identical; if therefore m be a constant multiplier, 
 A=ma, B=mb, C=me', A” =ma", B'’=mb", D=md. 
 
 But in order to determine the complete intersection of the 
 surfaces, we must combine their equations. ‘Therefore, mul- 
 tiplying the latter by m and subtracting it from the former 
 and having regard to the above relations among the co- 
 efficients, we find 
 
 (C—mc)2°+2(A'— ma) yz +2(B’— mb’) z7+2(C”’—me")z=0; 
 
 the equation to a surface which contains all the points com- 
 mon to the two proposed surfaces. But this may be decom- 
 posed into 
 
 %=0, (C—mc)x +2(A’— ma’)y+2(B'—mb')a +2(C”—mc")=0; 
 
 the first belongs to the assumed curve of intersection in the 
 plane of wy; the second is the equation to a plane, and 
 cannot therefore, when combined with either surface, give 
 any thing except a plane curve of the second order for the 
 other curve of intersection. 
 
SECTION III. 
 
 ON THE PROJECTIONS OF LINES AND PLANE SURFACES, 
 AND ON THE TRANSFORMATION OF CO-ORDINATES. 
 
 Tur meaning of the projection of a point, and of a line, 
 upon any plane has already been explained, (Arts. 2, and 13.) 
 Moreover if from the extremities of a limited line we drop 
 perpendiculars upon any indefinite line either in the same 
 plane with it or not, the part of the latter line intercepted 
 between the feet of the perpendiculars, is called the projection 
 of the former line upon the latter. 
 
 Also, if the sides of any plane surface be projected upon 
 a plane, the figure bounded by these projections is called the 
 projection of the given surface upon that plane. Between 
 the length of a finite line and the length of its projection 
 upon any plane or line, and also between the area of any 
 plane surface and the area of its projection upon a plane, a 
 remarkable relation exists, which we shall now exhibit. 
 
 79. The length of the projection of a limited line upon a 
 plane, is equal to the length of the line multiplied by the 
 cosine of the acute angle which it forms with the plane. 
 
 Let CD (fig. 25) be the line produced to meet Gd K, the 
 plane on which it is to be projected, in H. Let DHd be 
 the projecting plane, intersecting GdK in Hd; and in the 
 projecting plane draw Cc, Dd, perpendicular to Hd, and CD’ 
 parallel to it; then ed is the projection of CD, andz DHd =i 
 is the inclination of CD to the plane GdK; also CD’ = CD 
 
 xX COS 23 
 
 « ed= CD = CD cosi. 
 
63 
 
 80. The length of the projection of a limited line upon 
 any other line, is equal to the length of the line multiplied by 
 the cosine of the acute angle which the two lines form with 
 one another. 
 
 Let CD (fig. 26) be the line, 4B the line upon which 
 it is to be projected, Cc, Dd perpendiculars upon AB, then 
 ed is the projection of CD upon 4B. Let Ne, Md be planes 
 through C, D, perpendicular to AB, and of course containing 
 the lines Cc, Dd; draw CD" parallel to AB, meeting the 
 plane Md in D’, and join DD’; then the triangle CDD’ is 
 right-angled at D’, and DCD’ =i is equal to the angle 
 formed by the two lines, .. CD’ = CDcosi, but ced = CD’, 
 each being the perpendicular distance of parallel planes ; 
 
 *, ed = CD cosi. 
 
 81. ‘The area of the projection of any plane surface upon 
 a plane, is equal to the area of the surface multiplied by the 
 cosine of the acute angle which the planes form with one 
 another. 
 
 Let ACB (fig. 27) be a triangular area traced in a plane 
 which intersects the plane Gd, upon which the projection is 
 to be made, in GK. Through the angular points draw planes 
 AGa, BKb, CHc perpendicular to GA; then these planes 
 will contain the perpendiculars Aa, Bb, Cc let fall from the 
 angular points upon the plane of projection, and their inter- 
 sections with each of the planes 4H B, a Hb, will be perpen- 
 dicular to GK; also the angle CHe will equal the inclination 
 of the planes =7. Join the feet of the perpendiculars; then 
 acb is the projection of ACB; also join Dd, which is parallel 
 to Ce. 
 
 Then the triangles acd, ACD, since they have a common 
 altitude GH, are to one another as their bases ed, CD; simi- 
 larly the triangles bed, BCD, having a common altitude HK, 
 are to one another as cd to CD; 
 ed 
 CD? 
 = triangular area ABC, cos?, 
 
 " triangular area abe = triangular area ABC. 
 
64 
 
 Next suppose the figure to be projected is a polygon; then it 
 can be divided into triangles, each of which will be to its 
 projection as 1 to cosi; and therefore the sum of the triangles 
 will have to the sum of the projections the same ratio, or 
 
 area of projection = area of polygon . cosi ; 
 
 and as this is true however much the number of sides of the 
 polygon be increased, it is also true when the figure to be 
 projected is bounded by a curve ; 
 
 .". area of projection of any plane surface = area of surface. cos i. 
 
 82. The square of the area of any plane surface is equal 
 to the sum of the squares of the areas of its projections on 
 three co-ordinate rectangular planes. 
 
 Let the given area be denoted by 4, and its projections 
 on the planes of wy, yx, xu by A,, A,, A,, respectively. 
 Also let a, (3, yy denote the angles which a perpendicular to 
 the plane of the given area from the origin, makes with the 
 axes of w, y, 8. ‘Then vy (Art. 27) is the inclination of the 
 plane of the given area to the plane of wy, and therefore 
 A,= Acosy; similarly A, = dA cosa, A,= A cos DB; 
 
 “. 42 + Ai + At = AP $ (cos yy)? + (cosa)’ + (cos B)?} = A?. 
 Cor. Hence in fig. 9, since the triangles 4OB, BOC, 
 
 COA, are the three projections of the triangular area ABC, 
 if OA, OB, OC be denoted by a, 8, c, 
 
 (AABC)? = ($.4b)° + (f bc) + (hea)? ; 
 ». DABC =}/ (aby + (ao) + OO 
 
 Oblique co-ordinates. 
 
 83. To express the distance of a point from the origin 
 in terms of its oblique co-ordinates. 
 
 fig 2, AL the angles yOx, Ow, wOy, be denoted by 
 A; wy v3; and let OA, AN parallel to Oy, and NM parallel 
 to Ox, be the co-ordinates w, y, x of the point M. From 
 
65 
 
 the points 4 and N drop perpendiculars 4m, Nn upon Ox; 
 then mv is the projection of 4N upon Oz, and = AN cos), 
 since AW is parallel to Oy (Art. 80); and Om = O4 cosy. 
 Now from the triangle ONM, since MN is parallel to Ox, we 
 
 have 
 
 OM? = ON? + NM? +2MN.NO cos s ON. 
 
 But ON? =a? + y° +2vy cos v, from the triangle OAN, 
 and ON cosz ON = On = Om +mn = 4 cosun+Ycosr; 
 . P= a +y’ +2 + 2rycosy + 2a cose + 2yX COSA. 
 
 Cor. Since d is the diagonal of the parallelopiped of 
 which a, y, s are three conterminous edges, the above formula 
 gives the diagonal of any parallelopiped in terms of its edges 
 and the angles which they make with one another. It is also 
 the equation to the surface of a sphere whose center is at the 
 origin and radius = d. 
 
 84. To find the distance between any two points in 
 , terms of their oblique co-ordinates. 
 
 Let vw, y, x; @, y’, x’, denote the co-ordinates of two 
 points referred to oblique axes; then as in Art. 6, if through 
 these points we draw six planes parallel to the co-ordinate 
 planes, we shall form a parallelopiped with its edges parallel 
 to the axes, their lengths being a —a, y’—y, and 2 —3; 
 and the distance of the points is the diagonal of this 
 parallelopiped ; 
 
 “ @=(v — 2) + (y’ —y) + (2 - 2)? 
 +2(a' —x)(y’— y) cosy +2(a’ —x)(x'—2)cosu+2(y’—y) (2-2) cosa. 
 
 Cor. Hence we have the equation to the surface of a sphere 
 
 , 
 
 whose radius is d, and co-ordinates of its center, a’, y’, 3’. 
 
 85. To find the angle between two lines, whose equations 
 are given referred to oblique axes. 
 
 Piet 
 
 L=Mz) vx =mMsZ ; 
 , » be the equations to two lines 
 y=ns) yYEns 
 
 parallel to the proposed ones, through the origin. 
 
 4 
 
66 
 
 Then, as in Art. 30, if we take a point in each of these 
 lines, at a distance = 1 from the origin, and call their co-ordi- 
 nates v, y, 3 wv, y’, 2’, respectively; also their distance d, 
 and the angle between the lines 0, we have 
 
 2—2cos@ = d* = (a — x)? +(y'-y)* +(x’ —2)?+2(a'— x) (y'—y) cosp 
 +2 (« — x) (2-2) cosn + 2(y’— y) (2-2) cosa, 
 or, since | 
 l= a+ y? +2" + 2evy cosy +222 cosp + 2YX COSA; 
 
 1 
 
 wo? 4 y? + 2 4 20'y' cosy + 24a'2' cosm + 2y'X COSA, 
 cos 0 = wa! + yy + x2" + (ay + vy’) cosy + (a's + w2’) cosp 
 + (yz + 2’y) cosa. (1) 
 But since w=ms, y=ne, & =m’'s’, y =n'", 
 1=3°(1 +m’? + n° +2mn cosy +2mcosu+2nCcoSd), 
 1=2?(1 +m? 4? +2m'n' cosy +2m cosu + 2 cosa); 
 * cos@ = | 
 
 1+mm’+nn'+(m'n+mn’)cosv+(m+m’')cosn+(n+n’)cosr 
 
 rie 
 i 
 
 0 é 
 a/ 1+m?-+n®+2mn cos +2 m Cos 1+2n cosry/1+m”?+n?+2 m’n’cosv+2m’cospr+2 n’Cosr 
 
 Cor. Hence the condition that the lines may be at right 
 angles to one another, is cos @ = 0, or 
 
 1+mm' +nn'+ (m'n + mn’) cosv + (m +m’) cos u 
 
 +(n +n’) cosr = 0. 
 
 86. To find the conditions in order that a straight line 
 and a plane, referred to oblique axes, may be perpendicular 
 to one another. 
 
 et a line and plane be drawn through the origin, re- 
 spectively parallel to the proposed ones, and let their equa- 
 tions be 
 
 \, and s= Aw + By. 
 
 Y= NB 
 
67 
 
 Now this line, since it is perpendicular to the plane, is per- 
 pendicular to any line situated in the plane, and consequently 
 to the trace of the plane on wz, whose equations are 
 ae eel 
 9=—8, y=O; .*. (Art. 85) since m’=—, n =0, 
 
 A 
 
 ri ll i ( 7) + ent 
 —- —— COS MW —— We COS 7’ COS = 
 Fyne 4 vt+ uy, fs : 
 
 or A(1+mcosu+ncosr) + (m+ ncosy + cosp) = 0. 
 Similarly, 
 B(1+mcospu + 2 cosr) + (2 + mcosy + CosA) = 0. 
 
 Cor. Hence, we can find the angle between two planes 
 referred to oblique axes; for the above conditions enable us to 
 find the equations to two lines through the origin respectively 
 perpendicular to them; and knowing the equations to the 
 lines, we can compute the angle between them, that is, the 
 
 _angle between the planes, by Art. 85. 
 
 87. If we employ the symmetrical forms of the equations 
 to a straight line 
 
 we obtain immediately, for the angle 9 between these lines, 
 from equation (1), Art. 85, denoting by wy the angle between 
 the axes of x and y, and similarly of the others, 
 
 cos @ = Il’ + mm’ + nn' + (lm' + I'm) cos vy + (ln +1'n) cos vz 
 + (mn' + m’n) cos yx. 
 If we now make the former line to coincide successively 
 
 with the axes of a, y, x, and call the inclinations of the latter 
 
 line to those axes a, (, vy, we get 
 5—2 
 
68 
 
 , 
 cosa =l1' 4+ m'cosavy + n' cos az, 
 , , U 
 cos 3 = m' + I’ cosvy + n' cosy®, 
 cosy =n + I cos ax + m' cos ys; 
 -- cos 0 =l cosa + m cos 3 + n cosy, 
 
 where a, 3, y are the inclinations of one line to the co-ordinate 
 axes, and J, m, n the projecting ratios of the other. 
 
 Hence we obtain for the condition of two lines being 
 parallel, 
 
 lcosa + mcos 3 + n cosy = 13 
 and for the condition of their being perpendicular, 
 Icosa +m cos + ncosry = 0. 
 
 Likewise these equations express respectively the conditions 
 for a line whose equations are 
 OP PS 
 
 Lome ne 
 being perpendicular, and parallel, to a plane whose equation is 
 
 vw cosa + ycosP + % cosy = p. 
 
 Transformation of Co-ordinates. 
 
 The discussion of the nature and properties of surfaces 
 is much facilitated, when their equations are reduced to the 
 most simple form they are capable of, without being deprived 
 of any of their generality. This simplification is effected 
 by giving a suitable position to the origin, and suitable direc- 
 tions to the axes of the co-ordinates. We shall therefore next 
 proceed to investigate the chief formule for transformation of 
 co-ordinates. 
 
 88. To change the origin of the co-ordinates without 
 altering the directions of the axes. 
 
 Let a, y, x, be the co-ordinates of the point M (fig. 3) 
 referred to the origin O, and axes Ow, Oy, Oz; also let M' 
 
69 
 
 be the new origin, OA’=h, A’N’=k, N’M’=1, its co-ordinates, 
 and a’, y’, s, the co-ordinates of M referred to M’ as origin 
 and to axes parallel to the original ones; then _ 
 
 e= OA= MH + OA =a' +h, 
 similarly, y=Y th, sax 41; 
 
 and substituting these values of w, y, x in the equation to the 
 surface, we shall obtain the equation referred to the new origin 
 and to axes parallel to the original ones. 
 
 89. To pass from one system of co-ordinates to another 
 having the same origin, supposing the first rectangular, and 
 the second oblique. 
 
 Let Ow, Oy, Ox (fig. 28) be the rectangular, and O2’, Oy’, 
 Oz’ the oblique axes, having a common origin O. From any 
 point P draw PM, PM’, respectively parallel to Ox, Oz’, 
 meeting the planes of wy, a'y’ in M and M’, and draw MN, 
 M'N’ parallel to Oy, Oy’. 
 
 Then the co-ordinates of P in the two systems are 
 ON = aN Moen yy ALB i= 2, 
 ON’ =a, NM =y', MP=-2' 
 and our object is to express each of the first set in terms 
 of the latter. From N’, M’, drop perpendiculars V’n, M’m 
 upon Ow, and join PN which is also perpendicular to Ow; 
 
 then On, nm, mWN are the projections of ON’, N’M’, M’P 
 upon the axis of #; therefore (Art. 80) 
 
 , 
 On =x cosv# a, nm=y'cosya, mN =2' cosz'a, , 
 
 denoting, by wx, the ZO contained by the axes of «” and 
 #2 produced in the positive directions, and similarly of the 
 others. 
 
 . @=ON = Ont+nm+mN 
 
 = a cosa’« + y' cosy’a# + 2’ cosx’a@3 (1) 
 
 that is, each primitive co-ordinate is equal to the sum of the 
 projections of the three new co-ordinates upon its avis. 
 
70 
 
 Hence y= a'cosa’y + y cosy’'y +2 cosx’y, 
 / , , , , re 
 =a cosvs +y CosyX + COS8R; 
 or, if we denote by (m, m, 7) the cosines of the angles 
 which the axes of a’, y’, x’, make with the axis of w; and 
 by (m', n’, 17’), (m", n”, vr”) similar quantities relative to the 
 axes of y and z, we have 
 
 v=me+ny + re’, 
 
 y 
 sama +n'y + 7's". 
 
 Pee Lad 2 ‘Seth 
 ML+NY +7TR, 
 
 Of the nine angles involved in these formule, six only are 
 independent, there being three equations of condition; for 
 since wv, wy, a's, are the angles which a straight line, viz. 
 the axis of w’, makes with the three rectangular axes of 
 Uy Y, &; 
 
 .*. (cos aa)? + (cos ay)” + (cos az)? = 1; 
 and similarly with respect to the angles which the axes of 
 y and x’ make with the primitive axes; therefore 
 
 m +m? +m”? = 
 
 n+n'? +n’ 
 
 1, 
 BE AREDE 
 ye 2? 4 7’? = 1, 
 
 Cor. In the figure, the axes of a’, y’, x’ are supposed 
 to make acute angles with the axis of w; if this were not 
 the case, and if for instance one of them Oy’, made an 
 obtuse angle with Ow, m would fall to the left of », and 
 we should have 
 
 e®= On-nm+mN, 
 
 with which the general expression (1) still agrees, because 
 the term y'cosy'# is negative in the case supposed, but. 
 numerically equal to the projection of y’. If again, the 
 axes being as in the figure, y’ were negative, the point m 
 would fall to the left of , and we should have 
 
 e=On-nm+mNn. 
 
 Hence, we collect that the formula (1) is applicable to alk 
 
wil 
 
 cases, provided we pay attention to the signs of the co- 
 ordinates, and of the cosines; the angles, as was before 
 observed, being those formed by the axes produced in the 
 positive directions. 
 
 90. To pass from one system of rectangular co-ordi- 
 nates to another also rectangular. 
 
 Here it would be sufficient to join to the expressions, 
 and equations of condition of the preceding Art., three new 
 equations of condition expressing that the axes Oa’, Oy’, O3' 
 are at right angles to one another. 
 
 _- We may, however, arrive at the expressions for x, y, %, 
 briefly as follows, Let d be the distance of P from the origin ; 
 then referring the lines OP, Ow to the rectangular axes Oa’, 
 Oy’, Ox’, we have for the angle between them 
 
 , , 
 cos Px = cos Px’ cos x2’ + cos Py’ cos vy’ + cos PX cos #2; 
 
 hence, multiplying by d, and observing that dcos Pw = x, 
 dcos Pav’ = «’, &c., we have 
 
 , ? , , , , 
 V=H COSA’ + Y COSYXR4% Cosza, 
 and similarly for y and x; hence denoting the cosines as 
 before, 
 L=me +ny + 72, 
 y=ma + n'y + 1'2',70.(3)- 
 s= malt n'y +72". 
 In this case there are only three independent angles. For 
 since the axes of a and y’ are at right angles, 
 
 cos wa cos y’#@ + cos vy cos y’y + cos ax cos y/’% = 0; 
 similarly, expressing that the axes of w’, x’, and the axes of 
 y', %, are at right angles, and substituting for the cosines their 
 values, we have three new equations of condition to be joined 
 to those of the preceding Article, viz. 
 
 mn +m'n' + mn” = 0, 
 mr +m’ + my" = 0,>)...(4). 
 
 Ld a A 
 
 nr +-n7r +n’r”’ =0. 
 
72 
 
 91. Sometimes, in, the case of two rectangular systems, 
 it is required to find a’, y’, x in terms of a, y, x. This 
 may be effected by regarding a’, y’, 2’ as the primitive co- 
 ordinates, and recollecting that each is equal to the sum of 
 the projections of the co-ordinates 2, y, upon its axis; 
 
 “. @ = @ cosaxu’ + y cosya’ + 2% cosza’; 
 and similarly for y’ and x’; hence 
 
 v=maxe+my +m’, 
 ne + n'y + n's,>...(5). 
 v= rat ryt v's. 
 
 I 
 
 These expressions might also have been obtained from (3) by 
 adding them together after having multiplied them respectively 
 istly by m, m’, m”’; 2ndly by n, n’, n”; 3rdly by 7, 7’, 7; 
 reducing in each case by means of the equations of condition 
 
 (2) and (4). 
 
 Also the equations of condition, when «x, y, x are re- 
 garded as the new co-ordinates, and consequently their axes 
 referred to the axes of a’, y’, x, will be 
 
 m+n + 7. =1, 
 m”? +n”? + 7? =1,>...(6), 
 m+. 9? 4 9? = 1, 
 
 mm +-nn +r 
 
 0, 
 
 mm’ +nn" + rr” = 0,}...(7), 
 
 m'm" + n'n" + rr" = 0. 
 which are entirely equivalent to the relations between the same 
 constants obtained before, and may in every case replace them. 
 Indeed there is no difficulty in shewing that they are a neces- 
 
 sary consequence of (2) and (4). For since P is at the same 
 distance from the origin in either system, 
 
 (3) , Pe id 
 eP+y t+ Pax ty? t+ 2°75 
 
 and putting for «’, y’, x’ their values given in (5), and 
 equating coefficients on both sides, we obtain the equations of 
 condition (6) and (7). 
 
73 
 
 92. In the preceding Articles, the means of passing 
 from one rectangular system of co-ordinates to another, have 
 been given in simple and symmetrical formule; they have 
 however the inconvenience of involving nine constants, six 
 of which must be eliminated by means of six equations of 
 condition in order to make the determination of the new 
 axes relative to the primitive ones depend upon three quan- 
 tities. This led Euler to invent a mode of expressing each 
 of the nine constants in terms of three others, which are the 
 inclination of the planes of xy and x’y', and the angles 
 which their line of intersection forms with the axes of x and 
 
 , 
 
 x ; as described in the following proposition. 
 
 93. To pass from the rectangular system of co-ordinates 
 &, y, % to another rectangular system a’, y’, x’; having given 
 the angle @ at which the planes ay, «’y’, are inclined to one 
 
 another, and the angles @, Wy, which their line of intersection 
 makes with the axes of w and ’ respectively. 
 
 This is effected by passing successively through three 
 rectangular systems, each having one axis in common with 
 the preceding, and employing the formule relative to the 
 transformation of co-ordinates in one plane. 
 
 Thus to pass from the system of axes Ox, Oy, Oz, 
 (fig. 29) to the system Ow,, Oy,, Oz, of which Ov, is the 
 trace of the plane vw’ Oy’ upon wy, and Oy, is perpendicular to 
 Ow, in the plane of wy, we must put 
 
 @ = @,cos@ — y; sin a (1) 
 y= a, singd+ y, cosh 
 without altering x. 
 
 Next to pass from the system Ow,, Oy,, Ox, to the 
 system Ow,, Oy,, Ox’, of which Oy, lies in the plane a’y’, 
 and Oz’ is perpendicular to Oy, in the plane y,Oz, we must 
 put 
 
 Y,=y, cos 9 — 2’ sin :t @), 
 
 % = y, sin@ + 2’ cos@ 
 
 without altering 2, . 
 
44 
 
 Lastly, to pass from the system Ov,, Oy,, Ox’ to the 
 system Ow’, Oy’, Ox’, of which Oa’, Oy’, are at right angles 
 to one another in the plane x, Oy,, we must put 
 
 @, = # cosy — y' sin) 
 
 ’ - , apyd te} 
 
 Y2=@ sn + y cosy 
 without altering x’. But we may arrive at the result of 
 these three successive substitutions by a single substitution, 
 the formule for which will be formed by eliminating 2, y,, 
 and y, between the systems of equations (1), (2) and (3); we 
 shall then obtain for a, y, x in terms of a’, y’, 2’ and the 
 three angles, the following expressions : 
 
 w= sind sin @ + 2’ (cos@ cos yy — sin @ sin yy cos 8) 
 
 — y (cos d sin + sin d cos yp cos 6). 
 
 y= — x cos sind + 2 (cosy sind + sin yy cos d cos 8) 
 — y (sin ¢ sin yy — cos @ cos xf cos 8). 
 =x cos0 +a siny sin@ + y cosy sin 0. 
 
 94. If we dispense with the third transformation, we 
 shall have Ow,, one of the new axes, in the plane of xy. 
 The system Ow,, Oy,, Ox’, in which the new plane of wy 
 is inclined at an Z @ to the primitive one, and their inter- 
 
 section is the new axis of # inclined at an Z®@ to the pri- 
 mitive axis of w, is for most purposes sufficiently general. 
 
 The formule for it, combining the equations (1) and (2), 
 and calling the new co-ordinates w’, y’, x’, are ! 
 
 w= x cos — (y'cos@ — x’ sin @) sing, 
 
 y = sind + (y’ cos 0 — x’ sin 8) cos , 
 
 z=y sin@ +2’ cos@; 
 which evidently agree with the formule of Art. 93, when we 
 put WW =0. 
 
 95. In the preceding transformations we have supposed 
 the origin to remain unaltered ; if, however, the origin is to be 
 
45 
 
 changed, as well as the directions of the axes, we must employ 
 the formule 
 
 e=v' th, yay th, s=2" 4+), 
 
 where h, k, ¢ are the co-ordinates of the new origin parallel 
 to the primitive axes, and a”, y”, =” denote the values of 
 xv, y, = found in each of the preceding cases. 
 
 96. It is important to observe that, in employing any 
 of the preceding formule to refer an equation F(a, y, x) =0 
 to new axes, the transformed equation F'(a’, y’, x’) =0 will 
 always be of the same degree as the primitive equation ; 
 understanding by the degree of an equation the sum of the 
 indices of the three variables in that term where it is greatest. 
 For let this term be Aa? y! z'; then it becomes by substitution, 
 
 A (ma’ + ny + 73")? (ma! + n'y! + 9's')1 (ma + ny! + 7" 2’); 
 
 now these factors cannot furnish terms whose dimensions 
 exceed p, gq, t, respectively; therefore if A’a'?’y’Vs'" be 
 the term of highest dimension, p’ + q + ¢’ cannot exceed 
 p+q+t; ie. the dimension of the transformed equation 
 cannot be greater than that of the primitive. Neither can 
 it be less; for if it could, then, as transformation of co-ordi- 
 nates can never raise the degree of an equation, we could not 
 return from F(a’, y’, s')=0 to F(a, y, %)=0, which is 
 absurd. 5 
 
 97. To pass from one system of oblique co-ordinates to 
 another also oblique. 
 
 Let OM=2, MQ=y, QP=2x, (fig. 30) be the co-ordi- 
 nates of a point P parallel to the axes Ow, Oy, Ox; and 
 OM'=.2', M'Q’=y, QP=-2' the co-ordinates of the same 
 point parallel to the axes Ow’, Oy, Oz’. Through the 
 origin draw a normal ON” to the plane of wy, and on 
 the same side of it as the positive direction of the axis 
 of x; and from P, Q', M’ drop perpendiculars Pp, Q’n, 
 M'm, upon ON"; 
 
"6 
 
 then PQcos N”Ox = Op = Om+mn+np 
 = OM’ cos N" 2’ + M’Q’ cos N"y' + Q'P cos N"’2’, 
 or xcosN” 2 = 2’ cos N”’ a2’ + y’ cos. N"y' + x cos N’’2’. 
 
 Similarly, if we draw normals ON, ON’ to the planes of 
 yz, xx, and on the same side of them as the positive 
 directions of the axes of w and y, we shall have 
 
 «cos Na = a cos Na' + y' cos Ny’ + 2’ cos N2’, 
 y cos N’y = x cos N’2’ + y' cos Ny’ +’ cos N's 
 
 Since cos VW" x = sin (x, wy), &c. these formule may be 
 expressed without the aid of the auxiliary normals; but they 
 are more convenient in their present shape, as the employment 
 of negative angles is thereby avoided. 
 
 98. The position of a point P’ in space may be fixed 
 by the following three variables; (1) the radius vector OP’=r 
 (fig. 16); (2) the 2 COP’ = @ which the radius vector makes 
 with the axis of x; (3) the 2 AOA’ = q which the meridian 
 plane COP’ makes with the fixed plane COA. Of these 
 angles the former varies from 0 to 180°, and the latter from 0 
 to 360°, in order that the radius vector may pass through 
 all points of space. If #, y, x be the co-ordinates of P’, they 
 can be expressed by means of r, 0, x2 for P’M = r cos 0, 
 OM =r sin 0; 
 
 . w= OMcosh =r sinOcosd, y=rsinOsing, and x =rcos8; 
 
 and if these values be substituted for #, y, x in the rectangular 
 equation to any surface, we shall obtain the Polar equation to 
 the surface. 
 
 Plane sections of surfaces. 
 
 99. In the discussion of surfaces, it is useful to know 
 the nature and magnitude of the curves in which they are 
 intersected by any planes. To do this, it is not sufficient 
 to combine the equation to the surface F’ (a, y, ) = 0, with 
 the equation to the cutting plane z = dv + By +c, so as to 
 
77 
 
 eliminate one of the variables, x for instance; for the result 
 f(a, y,) = 0 would represent only the projection of the curve 
 required, which is not generally of the same nature with the 
 section in space, nor sufficient to determine it. But if we 
 transform the co-ordinates so that the cutting plane may be 
 that of w’y’, and then put x’ = 0 in the resulting equation, we 
 shall determine the trace of the surface on a’y’, i.e. the curve 
 in which it is intersected by the proposed plane. 
 
 And it may be here observed that, since the degree of an 
 equation is never altered by the transformation of co-ordinates, 
 if the equation to a surface be of the m' degree, the curve in 
 which it is intersected by a plane cannot be of a higher order 
 than the m™; but it may be of a lower if, by putting x’ = 0, 
 we cause all the terms of the highest dimension to disappear 
 from its equation. If we employ Euler’s formule for trans- 
 formation of co-ordinates (Art. 93), since x’ is to vanish in 
 the final result, we are at liberty to make z’ = 0 in the values 
 of x, y, x before we substitute them, and so we may obtain 
 the proper substitutions; these however, without being de- 
 
 duced from the general case, may be readily obtained by the 
 following independent method. 
 
 100. To determine the nature of the section of a surface 
 made by any plane passing through the origin. 
 
 The most convenient data for fixing the position of the 
 cutting plane are (1) the 20 at which it is inclined to the 
 plane of wy, and (2) the 2@ which its trace on that plane 
 makes with the axis of #; these are readily obtained if we 
 suppose the equation to the plane given; for-if the equa- 
 
 1 
 tion be s = Aw + By, then cos @ = tA, nad a Be (Art. 29), 
 
 A Tht : 
 and tan @ = — BR? Since Ax+ By =0 is the equation to the 
 trace. 
 
 Let «’ Oy’ (fig. 31) be the given plane, cutting the surface 
 in the curve 4M, and the plane of vy in the line Ow’, which 
 take for the axis of a’; and let Oy’, a line perpendicular to 
 
78 
 
 Ow’ in the given plane, be the axis of y’, and OR=2’, 
 RM=y’' the co-ordinates of any point M in the section re- 
 ferred to the axes Oa’, Oy’; also let OQ=a, QP=y, PM = 
 be the co-ordinates of the same point referred to the axes Oa, 
 Oy, Ox. Join RP, then 2 MRP =O the inclination of the 
 cutting plane to the plane of wy, and £wOza' = @ the angle 
 formed by its trace on wy with the axis of &. 
 
 Then PR =y' cos0, PM =~y'sin@; 
 OQ = OR cos g + RPsingd, QP= OR sin ® -hPcos@; 
 .e=y sind, v= cosp+y'cos6@ sin d, 
 y =w sind — y' cos 0 cos d; 
 and if these values be substituted in the equation F(a, y, s) =0 
 
 to the surface, we shall obtain a relation between a’ and y’ 
 which is the equation to the curve 4M. 
 
 101. If the cutting plane pass through one of the co- 
 ordinate axes, the formulee are simplified, and in many cases 
 are sufficiently general. 
 
 Let «’ Oy (fig. 32) be the cutting plane, passing through 
 the axis of y; Oa’ in the plane of za the axis of «’, 
 PM=2', OM =y, the co-ordinates of any point P in the 
 section, ON=a, NQ=y, QP==s the co-ordinates of the 
 same point. Join MQ, then 2PMQ= 90, and MQ =~2' cos@, 
 PQ =. sing; 
 
 . v= cos0, y=y, x= a' sind; 
 
 which shew that to determine the section made by a plane 
 through the axis of y and inclined at an 2@ to wy, we have 
 only to write w cos @ and 2’ sin@ for w and x in the equation 
 to the surface, without altering y; and the resulting relation 
 between w’ and y is the equation to the curve. Similarly, if 
 the cutting plane pass through the axes of s or w. 
 
 102. If the cutting plane does not pass through the 
 origin of the primitive co-ordinates, or if we wish to take a 
 
19 
 
 point in the cutting plane different from O for the origin of 
 the new co-ordinates, we shall only have to add to the second 
 members of the preceding formule the co-ordinates h, k, 1 of 
 the new origin reckoned parallel to the primitive axes. 
 
 103. To determine the nature of the curve formed by the 
 intersection of any plane with a surface of the second order. 
 
 Let Aa’ + By? + Cx* = 2A'w be the equation, which may 
 represent all surfaces of the second order; therefore, sub- 
 stituting for w, y, x the values found in Art. 100, 
 
 x= cosp+ y cos @sin @, 
 y = v' sin pd — y cos 0 cos , 
 s=y' sin 0, 
 the result developed and arranged will be 
 w (A cos’ p + Bsin’ p) + 2a'y' (A — B) cos Asin dh cos 
 + y” (A sin® d + B cos’ P) cos’ 6 + Csin’ 6} 
 = 24'u' cosh + 24’y' cos Osin d, 
 
 the equation to a curve of the second order, which will be 
 an ellipse, parabola, or hyperbola, according as the quantity 
 
 (A — B)’ (cos @ sin d cos p)? — (A cos’ @ + B sin? ) 
 x {(A sin’ + B cos’) cos’ 6 + C sin? 9}, 
 or — AB (cos 0)? — AC (cos ¢ sin 6)” — BC (sin @ sin 6)” 
 
 is negative, nothing, or positive. 
 
 Hence every section of an ellipsoid is an ellipse, because 
 all the quantities 4, B, C are positive. For a hyperboloid 
 of one or two sheets, in which cases one or two of the 
 quantities 4, B, C are negative, the section may be an ellipse, 
 parabola, or hyperbola. For paraboloids 4 =0; therefore 
 for the elliptic, in which case B and C have the same sign, 
 the section is an ellipse; except when 6=0, or @=0, in 
 which cases it is a parabola. For the hyperbolic paraboloid, 
 
80 
 
 since Band C are of contrary signs, the section is a hyper- 
 bola; except as before when 6 =0, or @=0, when it is a 
 
 parabola. 
 
 Cor. Since the section is referred to rectangular axes, it 
 can never be a circle unless the coefficient of #’y’ vanishes, or 
 
 (A — B) cos @ sin f cos h = 0; 
 
 Tw Tv ; 
 hence we must have ees or, pes or ¢=0; which 
 
 shew that for a circular section, the cutting plane must be 
 perpendicular to one of the principal planes of the surface. 
 In the next article we shall see that this property is confined 
 to one only of the principal planes of each surface. 
 
 104. Every surface of the second order, except the 
 hyperbolic paraboloid, may be generated in two ways by the 
 motion of a variable circle parallel to itself, the center of the 
 circle moving along a diameter of the surface. 
 
 Let the surface have a center, and let its equation be 
 Au’ + By’ + C2’ = D. 
 
 Since every circular section must be perpendicular to a 
 principal plane, let the cutting plane be perpendicular to va, 
 and inclined to wy at an Z @; and as the center of the circle 
 must be in the plane of za, let x=h, x= be its co- 
 ordinates ; then the equation to the section reckoned from that 
 point as origin will be 
 
 A(w cos0 +h)? + By’ + C(a’ sin 8 + 1)° = D, 
 (Art. 101) which represents a circle if 
 A cos’@ + Csin’@= B, or A + C tan’@ = B(1 + tan’), 
 B-A 
 i GER? 
 
 or tan? @ 
 
 and the relation between h and /, since the origin is the 
 center of the circle and therefore the coefficient of wv’ = 0, is 
 
 Acos@h4+ CsinOl=0; 
 
81 
 
 therefore the locus of the centers of the circular sections is 
 a straight line and a diameter of the surface. 
 
 ; : D Ah?+Cr 
 Also the radius of the section = Lot post BE Dh ae 
 B B 
 
 We must now examine, for each of the surfaces, which 
 axis it is that coincides with the axis of y to which the cutting 
 plane is parallel. 
 
 1 1 
 1. For the ellipsoid, A=, B==, C=5, 
 tan @=+-— 
 an @ 7 v= 
 
 -. b lies between a and c, or the axis of the surface to which 
 the cutting plane is parallel is its mean axis. 
 
 2. For the hyperboloid of one sheet, since we cannot 
 
 ; 1 ] 
 have B negative, we must put A=-, Bowe Gia aaa 
 
 i fe pes 
 a t Q = ck a CVE 3 
 an ayy 73 
 
 -.b>a, or the cutting plane is parallel to the greater of 
 the real axes. 
 
 3. For the hyperboloid of two sheets, since we cannot 
 have A and C negative, we must put 
 
 1 ] 
 
 A=, Bora, Coss 
 c G4 b* 
 
 anes NAS 9? 
 a 67 — Cc 
 
 ..b>c, or the cutting plane is parallel to the greater of 
 the imaginary axes. 
 
 Since tan@ has two values, the cutting plane may be 
 
 6 
 
82 
 
 inclined at an z@ or 180°—@ to the plane of wy, and hence 
 the surface may be generated by a variable circle in two 
 different ways; but it will be observed that in every case, 
 if the surface become one of revolution, the two positions 
 coincide in one which is parallel to the two equal axes. 
 
 105. Next, let the surface not have a center, and let 
 its equation be 
 
 By + Gz? =2A'x. 
 
 Then as before, if h, 7, denote the co-ordinates of the center 
 of the circular section, its equation reckoned from that point 
 as origin will be 
 
 By? + C (a’sin 0 + 1)? = 24'(a' cos + h), 
 with the conditions 
 
 B=Csin?@, Csin@.1= A’ cos@; 
 ‘sing = + JZ. and 1 =~ cot 8. 
 
 Hence the locus of the centers of the circular sections is a 
 straight line parallel to the axis of the surface, 7. e. it is a 
 diameter of the surface; also B and C have the same sign and 
 B<C; hence the surface is the elliptic paraboloid, and the 
 cutting plane is perpendicular to that principal section of the 
 surface whose latus rectum is the least. 
 
 In the case of the hyperbolic paraboloid, since B and C 
 have different signs, no plane can be drawn so as to intersect 
 it in a circle. 
 
 106. Any two circular sections of a surface of the second 
 order, provided they be not parallel to one another, are 
 situated on the same sphere, — 
 
 Since all circular sections are perpendicular to the same 
 principal plane, let (fig. 53) represent that principal plane, 
 
83 
 
 and let HA, E’B’ be the projections of any two circular 
 sections not parallel to one another, and also their diameters. 
 Bisect LA at right angles by JC, and let C be taken equidis- 
 tant from E and L’; therefore C is the center of a sphere 
 intersecting the proposed surface in the circle HA and passing 
 through £’, and therefore (Art. 78) intersecting the surface in 
 a plane curve, that is, in a second circle passing through LE’ ; 
 this circle must be either L’B’, or EA’ parallel to EA; but 
 it cannot be LA’, because two parallel circles on the same 
 sphere have their centers in a diameter perpendicular to their 
 planes, and here JOJ' being conjugate to the chords LA, E’A’ 
 cannot cut them at right angles unless the surface be one 
 of revolution, in which case 4’E’ and B'E’ become coincident ; 
 therefore the sphere will contain EB’. 
 
SECTION IV. 
 
 ON TANGENT PLANES AND NORMALS TO CURVE SURFACES, 
 AND THEIR VOLUMES AND AREAS. 
 
 107. To find the equation to the plane which touches 
 a given curve surface at a proposed point. 
 
 Let s=f(a#, y) be the equation to the surface, and 
 x, y, « the co-ordinates of the proposed point P (fig. 35) ; 
 then the equation to a plane passing through P will be 
 
 2s —-s=A(e —-a)+ By’ -y), 
 x’, y’, denoting the co-ordinates of any point in it. 
 
 Through P draw a plane parallel to z#, cutting the 
 surface in the curve PC, and the plane in the line PT'; 
 then these must touch one another at P, in order that the 
 plane may be the tangent plane to the surface at that 
 point. Now, for every point in the line PT’, y’ =y, and 
 therefore its equation deduced from the equation to the 
 plane is 
 
 x —s=A(a# — 2); 
 
 also the equation to the curve PC is deduced from x = f (a, y) 
 by making y constant in it; and in order that PT may 
 
 d :; 
 touch PC, we must have A=—, y being supposed con- 
 x 
 
 stant in the differentiation. Again, through P draw a plane 
 parallel to yz, cutting the surface in the curve PD and the 
 plane in the line PR; then, as before, the equation to PR is 
 
 s —-x= By -y); 
 
85 
 
 and the equation to PD is obtained from x = f (a, y) by 
 regarding # as constant ; and in order that PR may touch PD, 
 
 dz 
 we must have Ere being supposed constant in the 
 Ls 
 
 differentiation. Hence, substituting for A and B these values, 
 the equation to the tangent plane at a point a, y, 2, is 
 
 , 
 
 sey : ee ral ; 
 eae SS (VV — — : 
 i pares! y) 
 or, as it is usually written, 
 s—s=p(w™-—a)+q(y-y); 
 where p and q denote the partial differential coefficients of x 
 
 derived from the equation to the surface, and the co-ordinates 
 may be either rectangular or oblique. 
 
 Cor. 1. The plane whose position is thus determined by 
 the conditions that sections of it and of the surface, made by 
 planes parallel to two of the co-ordinate planes, touch one 
 another, contains, as we shall shew in the next Art., the 
 tangent lines of all curves that can be drawn on the surface 
 through the point in question. If the given equation to the 
 surface, instead of having the above explicit form, should be 
 u = F (x, y, 2) =0, then to determine p and q we have 
 
 de au . dus du 
 iieeoe Nady ads, 
 
 the differential coefficients being formed as if w, y, x were 
 independent ; hence, substituting for p and q their values, 
 the equation to the tangent plane under its most general 
 form is 
 
 du du du 
 , ae , ait o _ a O. 
 (v — a) one (y -y) a + (¢ — 2) FP 
 
 Corn, +2... TE ry denote the angle of inclination of the tan- 
 gent plane to that of wy, then (Art. 29) 
 
 1 
 cos y = ay aT ne 3? 
 1 ss ee 
 i | a 
 
86 
 
 and similarly, its inclinations to the other co-ordinate planes 
 may be determined. Also since the perpendicular on any plane 
 from the origin equals the constant term divided by the 
 square root of the sum of the squares of the coefficients of the 
 variables in its equation (Art. 9), the expression for the length 
 of the perpendicular on the tangent plane to a surface wu = 0 
 from the origin becomes 
 
 da Yay dz 
 
 NACE (S-) . ay 
 da dy ¥ dz 
 
 108. To investigate the equation to the tangent plane 
 at any point of a surface, considered as the locus of the 
 
 tangent lines to the surface at that point. 
 
 If a straight line be drawn through any point of a surface 
 and a contiguous point of the same, the limiting position 
 of this line as the latter point moves up to and ultimately 
 coincides with the former, is called a tangent line to the 
 surface at that point. The locus of the tangent lines at any 
 point of a curved surface is in general a plane, which is then 
 called a tangent plane at that point of the surface. 
 
 Let w=f (a, y, x) = 0, be the equation to the surface, 
 X-@v@ Y-y Z4-38 
 n 
 
 the equations to a straight line drawn through the point vyz 
 of the surface. ‘hen if this line meet the surface again in 
 the point a’y's’, 
 e=atir, y =y+mr, x =2z4+n7; 
 nes du , du du Q 
 Ae = wv & nea ® Tr +— Mr +-— NK + Ce 
 wp dex dy dx 
 du du 
 
 d 
 0 =/— + m— 
 oe da dy t dx 
 
 since f(x, y, #) the first term of the expansion of f(a’, y’, <’), 
 vanishes. 
 
 + terms in 7, (2) 
 
 eS 
 
87 
 
 ks e / , e e 
 
 Now let the point 2’y’s’ approach indefinitely close to 
 «yx whereby 7 is indefinitely diminished and the straight 
 line in question becomes a tangent line, then 
 
 du du du 
 
 Consequently from equations (1) we get 
 du du du 
 TAO) eee me) Viens — fe 
 be at) aig tole i veppity si Trai gs 
 
 a relation true for every tangent line to the surface at wysz. 
 And this is the equation to a plane; consequently it is the 
 tangent plane at the point wysx. 
 
 Cor. 1. In what precedes, we have supposed that 
 
 d d 
 Bo mies are finite, as they will in general be. If such a 
 
 dx’ dy’ dz 
 point be selected for the co-ordinates of which they all vanish, 
 the above equation gives no result; and the higher terms of the 
 expansion of f(a’, y’, 2’) must be employed. Taking the 
 terms of the second order in equation (2), dividing by 7? and 
 
 then making r = 0, we get 
 
 . at Be du Py Pu ne ad? u ay Pu 
 dx dy’ ds” dedy dads 
 +-2mn Pee = 0, 
 UL du 
 r i 2 
 on (4. — y)"= ph Borer 
 Oa sy (HREM hs ONGicice (Zi a) 
 
 9 
 
 du 
 + rat aia 
 
 4 
 
 the equation to the locus of the tangent lines, which is a 
 tangent cone, 
 
88 
 
 Points in a surface of this description are comprised 
 amongst what are called singular points ; the vertex of a cone, 
 or of a surface of revolution whose generating curve does not 
 cut the axis at right angles, are obvious examples; and we 
 observe that at such points the partial differential coefficients 
 al OS sean the Fortna( Chae 1 Art 107): 
 de dy 0 
 
 Cox. 2. Hence it follows that if a plane pass through a 
 generating line of a cylinder or cone, and through a tangent to 
 the base, and consequently through a tangent to every section 
 parallel to the base, it will be the tangent plane to the surface 
 at every point in the generating line; for it will contain the 
 tangents to all the curves traced on the surface through the 
 various points of the generating line. 
 
 109. To find the equations to the normal to a curve 
 surface at a proposed point. 
 
 The normal is a straight line drawn through the point of 
 contact perpendicular to the tangent plane; and will therefore 
 have equations of the form 
 
 we —-v=m(sx—s), y —y=n(z' -2), 
 xv, y, x being the co-ordinates of the point of contact; and 
 since it is perpendicular to the plane whose equation is 
 x —x=p(s'-a)+q(y'-y), 
 “. (Art. 25) m+p=0, n+q=0. 
 Hence, substituting for m and 7 their values, the equations to 
 the normal at a point wy, are 
 a’ —-a+p(s'—2)=0, y'-yt+q(s'-2) =0. 
 
 Cor. The angles a, 8, y formed by the normal with 
 the axes of wv, y, x produced in the positive directions, will 
 be given (Art. 28) by the formule 
 
 — Pp -4 
 cosa = : 9 COs SS 
 Jit p+ Vitp+¢e 
 1 
 cos’ = 
 
89 
 
 If the equation to the surface be given under the form 
 u = F (a, y, ) =0, substituting for p and q their values 
 (Cor. 1, Art. 107), we shall have 
 
 1 du B 1 du : 1 du 
 =—.—, cospP=—.—, co eas | ae 
 won he da R dy Y x? 
 du\* du\* du? 
 where R= () + ) a (=) 3 
 
 and the equations to the normal will assume the symmetrical 
 forms 
 
 dev dy dz 
 110. To find the length of the portion of the normal 
 
 intercepted between the surface and the plane of wy; and 
 the co-ordinates of the point where it meets that plane. 
 
 The co-ordinates of any point in the normal being 2’, y’, 2’, 
 and wv, y, x those of the point where it meets the surface, 
 the distance of these points is 
 
 V (w= 2) + (y= y) + (8) = (a) V1 peg. 
 But at the point where the normal meets the plane of zy, 
 % =0; .. the required length = — a/1 + p+ q’. 
 
 Also by making s’=0 in the equations to the normal, 
 we find 
 W=U+ ps, Y=yt qs, 
 the co-ordinates of the point where it meets the plane of wy; 
 
 and similarly the points where the normal meets the other 
 co-ordinate planes may be determined. 
 
 111. The equations to the normal may also be found 
 from the consideration that it is the longest or shortest 
 line which can be drawn from any point in itself to the 
 surface. 
 
 Let 2’, y’, x be co-ordinates of a fixed point in the 
 
90 
 
 normal, and a, y, x of any point in the surface; then if 
 © = distance of these points, 
 
 O° = (a - a)’ + (y'—y)? + (2-8), 
 a function of two variables w and y, for x is given in terms 
 
 of w and y by the equation to the surface. Therefore, in 
 order that 6? may be a maximum or minimum, 
 
 ass, d 
 a — © + —(s'— 2) =0, yy +7 (2) =0, 
 
 These are the equations which determine the position of the 
 longest or shortest line that can be drawn from the fixed point 
 w’y's to the surface, and consequently those to the normal. 
 
 112. To find the differential coefficient of the volume 
 of a solid bounded by the co-ordinate planes and_ planes 
 parallel to them, and by a given curve surface. 
 
 Let xs=f(#, y) be the equation to the surface, and 
 %, y, #% the co-ordinates of any point M (fig. 38) in it. 
 Through M draw planes parallel to the co-ordinate planes ; 
 then the volume V of the figure NMEF is a function of 
 wv and y; and if w and y receive increments h and k, the 
 point M will be brought to C, and the increment of the 
 solid contained by the planes ME, SD, SB, FM, will be 
 
 dV dV eh ONS ’V GV Fk 
 
 Be tay & eda © Gnay higye ys ees 
 But if we had made @ alone vary, the increment would 
 have been 
 
 dV a . 
 MBRF =—h-+ e, SLES, Fie 
 dav da’ 11.2 
 
 similarly, if we had made y alone vary, the increment would 
 have been | 
 
 TR hak Pai tl tH 
 
 MDEQ = —k + —, — 
 
 dy dy 1,2 
 
 + &e.; 
 
 therefore, subtracting, we have 
 2 
 
 | 7 hal 
 l.M = 
 vol. MCRQ adie 
 
 hk + &e., 
 
91 
 
 vol. UCRQ av 
 hk ~ dady 
 
 hence, taking the limit of both sides by making A and k& 
 vanish, in which case vol. MCRQ, becoming ultimately a 
 prism whose base is QR and altitude MP, = xhk, and all 
 the terms in the second member after the first, since they 
 involve different combinations of positive powers of h and k, 
 disappear, we have 
 
 or 
 
 + &¢.; 
 
 av 
 dady — es 
 
 113. Again, to find the differential coefficient of the area 
 of a curve surface, if S = area of the surface NM, by proceed- 
 ing as above we may shew that 
 
 2 
 
 surface MC = liu 
 dudy 
 
 surf. MC @S 
 
 a Ce 
 But surf. MC is ultimately coincident with the portion of the 
 tangent plane at M intercepted by the prism erected on the 
 base QR, and therefore = hk sec y, if sy denote the inclination 
 of the tangent plane at M to the plane of wy; 
 
 hk + &c., 
 
 and therefore limit of 
 
 ~ Ss Ws dz\? (dz\? i kop lie & 
 ee inp a + (=) a as ° ( rt. O7, or. 2). 
 
 114. In applying these results to find volumes and areas 
 of curve surfaces by double integration, considerable attention 
 is necessary in taking the integral between the proper limits. 
 This will be best seen by an example. Suppose it were 
 required to find the volume contained by two cylindrical 
 surfaces ferected on given bases ON, AN’ (fig. 37) in the 
 plane of wy whose equations are w= @(y), v= Wy (y)t, and 
 the curve surface BPP’ whose equation is z=f(#, y). In- 
 
 tegrating first with respect to #, considering y as constant, the 
 
 expressl ey 
 ession 
 
 P dudy 
 
 =2x=f (a, y), between the limits 
 
 w=GN=oly), ©=GN'=\-(y), 
 
92 
 
 we obtain a quantity, a function of y only, which when 
 multiplied by dy is the ultimate value of the slice PP’M 
 contained between two planes whose distances from za are 
 y and y+ oy respectively; and if this be now integrated 
 between the limits y= 6, y=b’, we shall have the volume 
 contained by two planes parallel to zw at distances therefrom 
 respectively equal to 6 and 6’. 
 
 115. Similarly, for the area of the surface, by integrating 
 the expression 
 
 &S WA i: ae (=) - F (2, y) 
 
 dady fk dx dy 
 
 with respect to x, considering y as constant, between the 
 limits v= GN, «= GN’, we obtain a quantity, a function 
 of y only, which when multiplied by dy is the ultimate 
 value of the portion of the surface PQ’ contained between 
 two planes parallel to xa at a distance dy from one another ; 
 and this integrated between the limits y= b, y =0’, gives us 
 the area of the surface intercepted by two planes parallel 
 to 2x. 
 
 116. To express the length of the perpendicular dropped 
 from the origin upon the tangent plane at any point of a sur- 
 face by polar co-ordinates. 
 
 Let ACB (fig. 41, bis) be the tangent plane at a point P 
 of a curve surface, CPF’ the tangent line to a section made by 
 a plane through the axis of z, DPE a tangent line to the 
 intersection with the surface of a right cone described about 
 CO with semi-vertical angle COP=6; GH the projection 
 of DE. ‘Then DOP is a tangent plane to the cone, and con- 
 sequently perpendicular to the meridian section of the cone 
 COP. Let OA=a, OH=a, OB=b, OG=0', OC H=c. 
 Then finding the equations to the traces OE, OD, we get the 
 equation to the plane DOE 
 
93 
 consequently the equations to a line through O perpendicular 
 
 to it are 
 x (- 7 Fd (; =) 
 e=-{|--—— = — |- : 
 GA Gliks seke 
 which line must lie in the plane COF whose equation is 
 
 ycos@ =wxsing, putting angle AOF = ¢; 
 : 1 1 1 1 
 sing (7 - 5) ~ 2089 (5-7) 
 (= @ cos 2) oe (= gm cos ) 
 we ry oy Bi 3 
 
 a b a 
 I cos @ sin ~\ 2 
 or ts (eee ae) 
 1 1 cos @ sin 2 2 
 = beers — > 6 ap wee ae J 1 
 a” +h b” ( a + b’ ( ) 
 
 Now let P = perpendicular on ABC from O, OP=r, OK =71', 
 p = perpendicular on PF’, p’ = perpendicular on GH; then 
 
 ] 1 1 1 1 
 =" > ary ea ees hae Is ar , 
 ee abide. if a bila 
 1 1 1 1 cos sin @\ ” 
 ay oe “2 ae Ww a a “+ (=e - ®) 3 
 eee OF RY ce a b 
 
 1 1 1 1 - @Q) 
 
 Now suppose r= f(@, 9) to be the polar equation to the 
 surface. Then, considering @ as constant, r =f (qd, @) is the 
 equation to the plane curve PM, and 
 lst) dine! ( sa) 
 
 p” gf °F yt dQ) ~ 
 
94 
 
 Again, considering @ as constant, 2 = sin Of (@, 0) is the 
 equation to the plane curve AL, and | 
 
 1 1 1 aa 1 1 A 1 eal 
 peat ee ~ sin? @ |r? ot dd ; 
 
 Hence, substituting in equation (2), we finally get 
 1 tv ft /dF\*> ‘cosec’’ Oo fare 
 RM avi 68 
 Peet r iy ao r da 
 an expression involving the partial differential coefficients of 
 
 r = f(@, 0), where @ and @ are independent of one another. 
 
 117. The above expression may of course be obtained 
 from the value of P given in Cor. 2, Art. 107, by changing 
 the independent variables. 
 
 118. In some cases it is convenient to express the dif- 
 ferential coefficient of the volume of a solid by the polar 
 co-ordinates explained in Art. 98. 
 
 Let ACP (fig. 39) be a curve surface; AP, A’P’, its 
 intersections with two conical surfaces described about OC 
 with semivertical angles 9 and 6+60; CP, CP’ its inter-— 
 sections with two planes through CO inclined to sa at 
 angles @ and @ + od; then it may be shewn as before, 
 if V = vol. ACPO, that 
 
 pyramid OPP = dV 
 See Se as 
 
 S00 dod 
 therefore, taking the limit of both sides by making ods O80, 
 vanish, in which case we may consider the base of the 
 pyramid as coincident with the surface of a sphere, center 
 O and radius OP, and therefore the area of the base will 
 =Pp.P'p=rsin0dd.r00, and the vol. of the pyramid 
 
 &e. ; 
 
 will =—.r?sin@d60, we have 
 3 
 
 da’ V 1 
 
 dOdp 8 
 
 which may be also deduced from the result of Art. 112, by 
 changing the independent variables. 
 
 r sin @; 
 
95 
 
 119. If in the above expression we substitute for r its 
 value in terms of @ and @, and integrate with respect to 0, 
 considering @ as constant, between the limits 
 
 9 =COG=F(¢), 0= COG = F,(), 
 
 (these values being obtained from the given equations to the 
 bounding conical surfaces BOD, B’OD’) we shall obtain a 
 quantity, a function of @ only, which when multiplied by op 
 is the ultimate value of the volume of the wedge GOH’ ; and 
 this, integrated between any two values of @ will give the 
 volume OBD' contained between two planes inclined at those 
 _ angles to xv, and between the given conical surfaces and the 
 curve surface. If instead of the vertex O, we suppose the 
 figure to be bounded by another curve surface whose equation 
 is x’ = f,(, 9), then the equation to be integrated will be 
 
 dv 
 
 120. ‘To express the differential coefficient of the area of 
 a curve surface by polar co-ordinates. 
 
 Proceeding as in Art. 118, it may be shewn that if S' be the 
 area of the surface APC (fig. bes we shall have ultimately 
 
 area of surface PP’ = 
 
 a 
 
 consequently, if p be the length of the perpendicular dropped 
 from O on the tangent plane at P, 
 
 vol. of pyramid OPP’ = 4p (er . ultimately, 
 
 d0do 
 also = 4” sin @. od 60, ultimately ; 
 
 iS sind ny 
 .' : ‘ A 6); 
 Dod =r A/ rsin'o (5) sint+ (— aa rt.116); 
 
 Pp 
 
 which may be also deduced from the result of Art. 113, by 
 changing the independent variables. 
 
SECTION V. 
 
 ON TANGENTS, AND NORMAL AND OSCULATING PLANES, TO 
 CURVES, AND THEIR LENGTHS. 
 
 121. ‘To shew how a curve in space may be represented 
 by equations. 
 
 We have seen (Art. 44) that the equation F' (a, y, x) =0 
 represents a surface, and if to this we join another equation 
 F’ (@, y, %) = 0 representing a second surface, and suppose 
 the variables to receive only such values as satisfy both 
 equations at the same time, we shall determine a series of 
 points situated in each of the surfaces, that is, in the curve of 
 their intersection. Conversely, as we have no other means of 
 determining a curve in space than by assigning two surfaces 
 each of which contains it, we cannot represent it analytically 
 except by two simultaneous equations among the variables 
 
 Vv, Ys 2. > 
 
 122, Among the various surfaces which may pass through 
 a curve and so determine it, we employ for the sake of sim- 
 plicity the cylindrical surfaces which are parallel to the co- 
 ordinate axes, as their equations will contain only two of the 
 
 variables. (Art. 43). 
 
 Let QPR (fig. 43) be a curve in space, and through all 
 its points draw perpendiculars Pp, Rr, &c. to the plane of 
 vy (or, if the axes be oblique, parallel to 4z); the assemblage 
 of these lines will form a cylindrical surface, called a projecting 
 cylinder of QR, and meeting the plane of xy in the curve qpr, 
 which is called the projection of QR on vy. Similarly, if we 
 drop perpendiculars from all the points of QR on yz and za, 
 we shall have two other projecting cylinders, and two other 
 projections ; and the curve will evidently be determined if any 
 two of its projections are given, for in that case two cylindrical 
 surfaces will be given of which it is the intersection. Now 
 
97 
 
 the projections qpr, Q'P’R’ are determined by equations of 
 the form 
 
 gi (av, y) = 0 ory=@ (2), Vi (8 &) = 0 or ¥ = Wy («), 
 
 the complete signification of which, as we know, is the cy- 
 lindrical surface erected upon each as its base. Therefore 
 the curve QPR will be determined by the system of simul- 
 taneous equations 
 
 y= (x), x=W(a). 
 
 In these, only one of the variables, w for example, is arbitrary ; 
 and any assumed value of w joined to the corresponding 
 values of y and x derived from them, will belong to a point 
 in the curve. The equation to the projection of QPR on the 
 plane of yz, is deduced from the other two by eliminating 2. 
 
 Hence it appears that every curve in space may be con- 
 sidered as formed by the intersection of two cylindrical surfaces 
 erected on its projections as their bases, perpendicular to the 
 co-ordinate planes. | 
 
 123. To find the equations to the line which touches a 
 given curve at a proposed point. 
 
 Let PT’ (fig. 43) be a tangent to the curve QPR at a 
 point P; also let qpr be the projection of the curve on the 
 plane of wy, and pé a tangent to the projection at p; we 
 must first prove that p¢ is the projection of PT. Let R be 
 a point in the curve near to P, r its projection; draw the 
 lines PR, pr; then pr is the projection of PR, and continues 
 so, however near & approaches to P, and consequently r 
 (which is always in a line through # parallel to Pp) to p. 
 ‘Therefore in the ultimate positions of the lines, when R co- 
 incides with P and r with p, and they become tangents at P 
 and p, pt is the projection of P7'; that is, the projection of 
 the tangent at any point coincides with the line touching the 
 projection of the curve at the corresponding point. The same 
 is of course true relative to the other co-ordinate planes; if 
 therefore y = d (vw), s = (a) be the equations to the pro- 
 
98 
 
 jections qr, Q’R’, the equations to the lines pt, PT’, or to 
 the line P7' of which they are the projections, will be 
 dz 
 
 d 
 s (w’ —#), x -—s= oT v— x); 
 
 y -y= 
 
 v, y, % being the co-ordinates of the point of contact, and 
 x, y’, x those of any point in the tangent, and the co- 
 ordinates being either rectangular or oblique. 
 
 124. The two equations to the curve may evidently be 
 supposed to arise from the elimination of ¢ between equations 
 of the form « = f(t), y=fi (4), 2 = fo (t); consequently, when 
 the co-ordinates of a point in the curve are all regarded as 
 functions of the same variable ¢, the equations to the tangent 
 line will take the symmetrical forms 
 
 dz dy dg 
 dt dt dt 
 
 Or if the equations to the curve, instead of having the 
 explicit forms assumed in Art, 123, should be 
 
 Mie VAG YS) = 0, We, et, (@, Ye) 0, 
 then considering y and x each as a function of wv, we have 
 
 du Hi dudy dwdz _ 
 dw dydzx A dzduv 
 du, du,dy du,dz 
 
 aye Ce a 
 
 d ’ : 
 from which =; pe may be obtained and substituted in the 
 v dx 
 
 equations to the tangent line. 
 
 b) 
 
 e 
 9 
 
 125. If the tangent to a curve makes angles a, (3, y 
 with the axes of aw, y, #, since its equations, taking the pro- 
 jections on the planes of x” and yx, are 
 
99 
 
 dy 
 , 1 , , da i 
 eae: Fa ©) Vom Ue (Fae ®)s 
 dx dx 
 dy 
 1 dav 
 cosa LS eee 
 Vise (Ge) Me) +() 
 a dx} ' \dwx i dx +(= 
 dx 
 COS? = 
 1+ (2) +(Z) 
 4 dx dx 
 
 126. To find the equation to the normal plane to a 
 curve at a proposed point. 
 
 A curve can have only one tangent at a proposed point, 
 but it may have an infinite number of normals, that is, of 
 lines perpendicular to the tangent through the point of con- 
 tact ; these all lie in one plane called the normal plane. Let 
 x, y, be the co-ordinates of the proposed point of the curve, 
 ‘then since the normal plane passes through that point, its 
 equation will be 
 
 vs’ —-2=A(e'-a)+ By -y); 
 
 and since it is perpendicular to the tangent whose equa- 
 tions are 
 
 dy 
 1 , , d. 
 x am (s — 2), Yonsself amar (Ham 8)s 
 da dx 
 dy 
 i dz 
 A+=—==0 =ume = 0) ; 
 Pr 4 HER ees 
 da dx 
 
100 
 
 hence, substituting for 4 and B these values, the equation to 
 the normal plane at a point wy%, is 
 
 : , dy ? dz 
 v—at+y FU) et TAS sn ti Baim 
 
 Cor. If a, y, x be each regarded as a function of the 
 same variable ¢, as in Art. 124, the equation to the normal 
 plane takes the symmetrical form 
 
 dx dy dz 
 a aki by inks Vee N iseadeeeiye 
 Dre) fay gl ee ey pinto Pa) tay 
 
 127. To find the equation to the osculating plane to a 
 curve at a proposed point. 
 
 The osculating plane at any point is that which has a 
 closer contact with the curve than any other plane passing 
 through the same point. 
 
 Let «#, y, x be the co-ordinates of the point of the curve ; 
 then the equation to a plane passing through it will be 
 
 ACK Saye BY Syl Ze oF =) 
 
 let Oo be the length of a perpendicular let fall on this plane 
 
 from a contiguous point in the curve whose co-ordinates are 
 
 wth, y+kh, x4+1; 
 Ah Bk ail 
 \/ A + B 4 C 
 dx dy ‘dz ad’ x CY 6a 
 
 pelea ty a cS) +}3( 4 Bey cS) ae 
 
 ( Ai cdi) \ al li, Ge en 
 
 tp (Cor. Art. 37). 
 
 3 
 
 if we suppose the points consecutive, and substitute for h, k, 
 l their developements in series ascending by powers of 7; 
 + being the increment of ¢ the variable of which the three 
 co-ordinates are assumed to be given functions. Now if we 
 determine the constants so that the coefficients of + and 7* 
 
101 
 
 may vanish in the numerator of 6, the portion of the curve 
 immediately contiguous to the point wysz will coincide more 
 nearly with the plane so determined than with any other 
 plane that can be drawn through that point; and as only the 
 ratios of A, B, C to one another are required, this can be 
 effected by establishing the two conditions 
 
 dx dy dz dx Ei d’ 
 — a —— =0; 4A — =| 
 Bri aeat) satin’ ae dae ede eae 
 
 whence eliminating A, B, C between these and equation (1), 
 
 2) 
 2 
 &e., by a’, 2”, &c., we get 
 
 di 
 and denoting a EE 
 
 (y/'2"—2!y")(X—a) + (w/a""—a'e”)(Y-y) +(a'y"-y'a")(Z—2) =0, 
 
 the equation to the osculating plane at a point wyz. 
 
 Cor. If we assume y and g to be functions of a, in the 
 ordinary way, then 
 
 d & d & 
 Ah+B/( ella ct 1) 40(4= + 4o + Be) 
 s_ dx dx dx dx” (2) 
 / A+ B+ C 
 dy d’y ad? z 
 oA — t= Oy ——- — = 
 eer cS aE ere et 
 
 and the equation to the osculating plane becomes 
 d’y Gydzs dzdy 
 oa rere ora 
 
 which evidently agrees with the above, when & = ¢. 
 
 as 
 — a)+ agit? - Y), 
 
 128. In order that the contact may be of the third order, 
 the coefficient of h* in the numerator of 6 in equation (2) 
 d? zx d’y 
 
 must also vanish, or Cis + Lister =e 
 
ae eT SE A eT 
 dx’ da da’ da’ : 
 the condition which must be satisfied at any point of the 
 curve where the contact is of the third order. When the 
 contact is only of the second order, 
 
 1 We Bs _ dy 
 bee ty" ily Ro ge 
 J A? + B+ C {3 (Cas ss ia) ate 4 } 
 
 a quantity which changes its sign at the same time that h 
 does ; consequently the curve generally cuts its osculating 
 plane. On the contrary, when the contact is of the third 
 order, the sign of 6 does not change with that of h; and 
 the curve is said in such cases to have a point of inflexion. 
 The condition then of there being such a point is 
 
 asdy ay ds 
 
 eS ea ee SSS 
 — _— 
 
 da’ dx*® da? dx’ 
 
 129. When a curve in space is a plane curve, if w, y, = 
 be the co-ordinates of any point in it, they will always satisfy 
 the equation to a plane 
 
 v= Aw + By’ +ce...(1), so thats = d~e+By+e; 
 
 therefore the differential coefficients of = and y will be such 
 as to satisfy the equations 
 
 dz iy dy ds d°y 
 
 a 
 dv dx dx dx®’ 
 
 A and B denoting constant quantities ; 
 
 ad’ 2 
 d [ dx’ Tyds asd'y 
 ed a\ ai) Esai a Me eae ee 
 dx* 
 
 which must be rendered identically true by the equations 
 y=@(v), x= \(v), when they represent a plane curve; 
 
103 
 
 otherwise, the curve is of double curvature. This condition 
 being satisfied, if we substitute for 4A, B, ¢ their values in 
 (1), we find the equation to the plane in which the curve is 
 situated the same as that to the osculating plane, as might 
 have been foreseen; for when a curve in space is a plane 
 curve, the plane in which it is situated is the osculating plane 
 at every point. Hence when the above condition is satisfied, 
 the equation to the plane in which the curve is situated may 
 be obtained by writing for the differential coefficients their 
 values in the equation to the osculating plane. 
 
 If we assume each of the co-ordinates vw, y, x to be a 
 function of the same variable ¢, a symmetrical expression for 
 the condition (2) may be easily obtained by eliminating the 
 constants from the three derived equations of lw + my+nzx=c. 
 
 130. The osculating plane at any point of a curve in 
 space is perpendicular to the line of intersection of con- 
 secutive normal planes at that point. 
 
 The equation to the normal plane at a point wyz is 
 
 dy sy ds 
 ve — Ut CRE mae 28D) Seye & E 
 
 : ; dy dz 
 and at the contiguous point 7+h, y+ he +&c., 2 the + &e. 
 
 it 1s 
 ’ dy dy d*y 
 wa -he(y -y- hg te.) (+ + hot 4 &e.) 
 
 d dz d® 
 + (¥ ~# ~h— — &e.] (= + a + Be.) = 05 
 
 or, combining it with the former and dividing by 4, it is 
 
 ; d’y dy d’s dz\? s 
 Bet ET ge (3) + apn (ag) ttm 
 h, h*, &c. = O. 
 
104 
 
 Therefore, making 4 = 0, the equations to the line of inter- 
 section of consecutive normal planes are 
 
 OY dz 
 w-—at+y a dean ne my eye 
 2 
 
 2 2 2 
 oF peo Cees) be (=) 5 (=) hee Os 
 dx” d x” dx dx 
 (the latter equation being evidently that which results from 
 differentiating the former with respect to #, considering «’, y’, 
 %’ as constant) and if we deduce the equations to the pro- 
 jections of this line on the planes of yz and ga, it will be 
 seen that the conditions of being perpendicular to the os- 
 
 culating plane (Art. 25) are fulfilled. 
 
 131. To find the differential coefficient of the length of 
 the arc of a curve in space. 
 
 Let y= (2), * =wW («) be the equations to the curve, 
 
 2, y, % the co-ordinates of any point P in it (fig. 43), 
 at+h,y+hk, 2+, the co-ordinates of a contiguous point R, 
 s =length of arc QP, Q being a given fixed point in the curve, 
 
 and &R = chord PR; 
 then R?=7? +447 
 
 | dy hd’ dz hh dx 
 ans (A 4 = <7 4 &e V4 + (Aa at &e.] 
 
 da rd dx da 2 dx dx’ 
 dy\? dz\* 
 = h? {1 a : yes 
 + (2) + oy fan + &c 
 DSN He... R a/ dy. * dz\? 
 , — = | fete — —— ae 
 ay Imit o h 1 + Gy ae (= 
 
 Cor. Hence the formule for the angles a, 3, yy, which | 
 
 a tangent at the point wys forms with the axes of a, y, 2, 
 may be written 
 
dy dz 
 dax da 
 cosa=—, cosB=——, cosy =—; 
 dw dv dn 
 or, supposing s to be the independent variable, 
 
 dxv B dy dz 
 a oe a eS a) co = —— 
 COAG = ——s COs rF sty 
 
 132. If we employ polar co-ordinates so that x =r cos 0, 
 # =rsinOcosd@, y =17 sin @sin @, then 
 
 d ; dr\* : d d\? 
 
 ie / + (=) + r’sin® @ (52) : 
 ds r 
 
 Also, —- = ————, p being the perpendicular on the 
 dr re p 
 
 tangent line at the extremity of r. 
 
SECTION VI. 
 
 ON THE DISCUSSION OF THE GENERAL EQUATION OF THE . 
 SECOND ORDER. 
 
 133. To find the position of the center of any surface. 
 
 The center of a surface is a point O (fig. 22) such that 
 any chord of the surface PP’, drawn through it, is bisected in 
 it. (It must be observed, however, that if PP’ cut the surface 
 in more points than two, it would be sufficient that these 
 points combined in a certain order should be, two and two, 
 
 equally distant from OQ). 
 
 If the surface be referred to any three axes originating 
 in O, and PM, P’M’ be the ordinates parallel to Ox of the 
 extremities of a chord, we see from the equal triangles POM, 
 P’OM’, that these ordinates are equal and of contrary signs ; 
 the same thing would be true for the other co-ordinates of P 
 and P’, as well as for every other chord passing through O. 
 If therefore f (xv, y, x) =0 be the equation to the surface, 
 and if it be satisfied by 7 =a, y= 6b, x =c, it must also be 
 satisfied by # = —a, y= —b, s = —¢; that is, it must be such 
 as not to alter when the signs of the three variables are 
 changed ; and, conversely, if it have this property, the origin 
 is the center of the surface. When f(a, y, %) = 0 is algebraic, 
 it cannot have the above property unless the dimension of 
 every term be even in an equation of an even degree, and the 
 dimension of every term be odd in an equation of an odd 
 degree; for in the former case the equation is not at all 
 altered by replacing vw, y, s by — a, — y, — #3; and in the latter 
 case (in which the equation cannot have a constant term) the 
 sign of every term will be altered, and therefore the whole 
 equation unaltered. 
 
107 
 
 Hence, to find whether a proposed surface admits of a 
 center, we must refer it to parallel axes, through a new origin 
 having co-ordinates h, k, 1, by putting 
 
 eaa't+th, yay +k, x= +], 
 and equate to nothing the coefficients of all the terms which 
 are of a dimension different (as far as regards odd and even) 
 from the degree of the equation; if these conditions can all be 
 satisfied by real and finite values of h, k, 1, the surface has 
 a center, and h, k, / are its co-ordinates; in the contrary case 
 the surface has no center. 
 
 134. To find the co-ordinates of the center of a surface 
 of the second order represented by the general equation of 
 the second degree. 
 
 The general equation of the second degree is 
 FS (&, Ys 8) = an? + by? +02? + 2a yx +20 su+ 2c uy 
 + 2a" e2+2b"y + 2c"s+d=0. 
 
 Here we must make v=a’ +h, y=y +k, s=2'+1, and 
 equate to nothing the coefficients of terms of odd dimensions, 
 that is, those involving a’, y’, 3’. The result of these sub- 
 stitutions is 
 
 S(@ +h, y +k, 2 +0, 
 
 which when expanded will consist of three parts, (1) terms 
 of two dimensions which must be the same as those of 
 FT (2, y’, #’) which are of two dimensions; (2) terms of one 
 dimension which are 
 
 va f(y i). df (h, bl). dfeecan 
 CMMNM ied aa) as Stas 
 
 each of which must disappear; and (3) a constant term 
 f(h, k, 1); therefore the result will be 
 
 aw” + by” +62" 420'y'x' + 2b sv + 2c uy +f (h, kh, D =% 
 
108 
 
 df (hs k; !) 
 dh 
 
 with the conditions =0, &c., or 
 
 ah+bl+ck+a"=0 
 bk + al+céh +h’ =0 wl)» 
 cl+ak+bh+c’ =0 
 
 for determining h, k, 7. Multiplying the two latter equations 
 respectively by indeterminate coefficients ¢, uw, and adding 
 them to the former, we have 
 
 (a+te +ub)h+a"+tb’ + uc’ =0, 
 provided c +tb+ua'=0, 6 +ta'+uc=0; 
 these two latter equations give ¢ and w, and then substituting 
 
 in the first we find h = = where 
 
 D=aa’ + bb” 4+ cc” —abe -2a'l'e," 
 N=a’ (bc — a”) +b" (vv — cc) +0" (ac — bb); 
 aN Wve 
 similarly, & = —, / = _. 
 y> D’ D 
 Hence, provided D is not = 0, these values of h, k, 1, which 
 are always real, are finite; and the surface will have a 
 single center of which they are the co-ordinates, and its 
 equation reckoned from the center as origin will be, suppressing 
 accents, 
 
 ax’ + by? +cex +2a ys + 2b ea4+2ay+ah+b'k 
 aa cl+d= 0% 
 
 for, multiplying equations (1) respectively by h, k, 7, and 
 adding, we have | 
 
 fi, ky Dl) = ah WK +014 d. 
 
 135. If the constant term f(h, k,l) disappears, the 
 surface represented is a cone, since if we combine its equation 
 with the equation to a plane through the origin s = 4a + By, 
 
109 
 & 
 the result will be of the form y = a (p + /q) indicating two 
 straight lines through the origin; unless the radical be 
 impossible for all values of 4d and B, in which case the pro- 
 posed equation represents a point. 
 
 If the coefficients of the given equation be such that 
 D=0, and the three numerators are not all =0 at the 
 same time, then one at least of the co-ordinates of the 
 center will be infinite, which signifies that the surface has 
 no center. 
 
 If at the same time that D =0 the three numerators 
 vanish, then the surface admits of an infinite number of 
 centers; for in that case the three equations (1) are reduced 
 to one, or to two realiy distinct equations, as is shewn in most 
 treatises on Algebra, and may therefore be satisfied by an 
 infinite number of values of wv, y, x. If they are reduced to 
 two, that is, if the values of h and k deduced from the two 
 first for instance, satisfy the third whatever / be, then there 
 will be an infinite number of centers situated in a straight line 
 which is the locus of the two independent equations; the 
 surface will therefore be a cylinder on an elliptic or hyperbolic 
 base. 
 
 If the three equations (1) are reduced to a single equation, 
 that is, if the value of # deduced from the first, for instance, 
 satisfies the other two whatever k& and 7 be, there will be an 
 infinite number of centers situated in a plane which is the 
 locus of the single independent equation, and the proposed 
 surface will be a system of two planes parallel and equidistant 
 from that plane. In this latter case the proposed equation 
 must be capable of being resolved into two rational factors of 
 the first degree. 
 
 136. The locus of the middle points of a system of 
 parallel chords of any proposed surface is called its dia- 
 metral surface. ‘This surface will have several sheets, if 
 each of the chords has more than two points in common 
 with the proposed surface; if, for instance, the proposed 
 surface be of the m™ order, the points of intersection with 
 
110 
 
 its chords, real or imaginary, will be in number m, and 
 their combination on the same indefinite line will form 
 in(n-—1) different chords, and as many middle points; 
 and therefore the diametral surface, since it may be met 
 by an indefinite line in $2 (2-1) points, will have an 
 equation of the degree $(m—1). For surfaces of the 
 second order where n= 2, the diametral surfaces can only 
 
 be planes. 
 
 When any surface admits of a diametral plane, if we 
 make it the plane of wy, and take the axis of x parallel 
 to the chords which it bisects, the equation to the surface, 
 for every pair of values vw =a, y = 6, must furnish for x 
 values which, taken two and two, are equal and of contrary 
 signs; and therefore the equation, supposed algebraic, can 
 only involve even powers of x. And, conversely, whenever 
 an equation contains only even powers of one of the variables, 
 x for instance, the plane of wy is a diametral plane, and 
 is said to be conjugate to the chords parallel to the axis 
 of zg. Also if a diametral plane be perpendicular to the 
 chords which it bisects, it is called a principal plane, and the 
 chords principal chords. Moreover the intersection of any 
 two diametral planes is called a diameter of the surface. 
 
 137. To find the equation to a diametral plane of a 
 surface of the second order. 
 
 Let v=mz, y=ns, be the given equations to a line 
 through the origin to which the proposed system of chords 
 is parallel, and f(a, y, 3) =0, the general equation of the 
 second degree, the equation to the surface. Let h, k, 1, be 
 the co-ordinates of the middle point of any chord, and let 
 the surface be referred to axes parallel to the former ones 
 passing through it; then the equation will become 
 
 S(a@ +h, oy +k, x +20) =0, 
 
 and the equations to the chord itself will be a = mz’, 
 y’ =n; therefore the values of x’ belonging to the points 
 
111 
 
 of the surface where the chord meets it, are given by the 
 equation 
 
 f(mx' +h, nz’ +k, x +1) =0......(1), 
 which is of the form 
 Rz? + S2'+T=0; 
 
 and since the values of x’ are equal and of opposite signs, 
 S=0. But Sis the coefficient of x’ in equation (1) ; 
 
 mdf (hy ky 1). df (hy yD af (hy hy 2) 
 dh eens S/%) Mark Soni lee Whe uae 
 
 or mM(ah+Ulock+a")+n(bk+al+ch +b") 
 tel+avk+h +c" =0...... (2); 
 or (am+cen+b)h+(bn+em+a)k+(c+0m+an)l 
 +am+b’n+c" =0, 
 
 the relation among the co-ordinates of the middle point of 
 any chord, or the equation to a diametral plane. 
 
 Cor. As the coefficients of h, k, J are possible, there will 
 be a diametral plane for all values of the constants m and n, 
 unless the three coefficients should all become nothing at the 
 same time, when the diametral plane will be situated at an 
 infinite distance. But the equations 
 
 am+cn+b =0, bn+cém+a=0, c+bm+an=0, 
 
 containing only two unknown quantities, have an equation 
 of condition which is the same as D = 0 (Art. 134); in this 
 case therefore the surface has not a center. 
 
 When the surface has a center, every diametral plane 
 passes through it, or through the locus of the centers; for 
 equation (2) is visibly satisfied by the co-ordinates of the 
 center furnished by equations (1), (Art. 134). This also 
 follows from the definition. 
 
112 
 
 138. Any diametral plane of a surface having a center 
 is parallel to the tangent plane applied at the extremity of 
 the diameter to which it is conjugate. 
 
 Taking the center for origin, the equation to the surface 
 
 will be 
 ax’ + by? + cx? +20 yx 4+ 202742 xy+d=0; 
 
 therefore the equation to the tangent plane at a point vyz is 
 (Cor. 1, Art. 107), 
 
 (av + bx +c'y) (# -— 2) + (byt+ax4+ec'x) (y’-y) 
 + (cx +a'y4+0'2) (x — 2) =0, 
 or (av +Uxtcy)a+(bytaxstea)y+(extay+d'x)2 
 +d=0; 
 
 and if it be applied at the extremity of the diameter whose 
 equations are v = mz, y = nz, the equation becomes 
 
 / ] , / 4 / , d 
 (am+cen+0')a'+(bnt+em+a')y + (c+ bm +a‘n)z'+-=0, 
 R 
 
 and therefore (Art. 18) represents a plane parallel to the 
 diametral plane which is conjugate to the diameter x = mz, 
 y= nz, the equation to which is (putting a” = 6b” =c”’ =0, 
 in the equation Art. 137) 
 
 (am+cn+)a + (bn+emta)y+(ce+ Um+a'n)x'=0. 
 
 This result might have been foreseen; because the straight 
 lines in which a diametral plane and a tangent plane at the 
 extremity of the conjugate diameter are cut by any plane 
 through that diameter, must, by the nature of lines of the 
 second order, be parallel to one another. 
 
 139. We shall now proceed to the reduction of the 
 general equation of the second degree 
 
 ax? +by+ex°+2a'yxt2bear2cuvy+2ax42b"y4 2c x+d=0, 
 
f 113 
 
 where we suppose the co-ordinates rectangular; for if they 
 were oblique, by transforming them to rectangular co-ordinates 
 we should obtain an equation of the same degree as the 
 above (Art. 96), and which could not therefore be more 
 general than the one which we have assumed. We shall 
 prove, as affirmed at Art. 53, that this equation, after being 
 simplified as much as possible, will always assume one or 
 other of the forms 
 
 Ax’ + By? + Cz’ =D, 
 By? + C2? =2A'e, 
 
 the co-ordinates being rectangular; and therefore the general 
 equation of the second degree can never represent any other 
 surface than one of those discussed in Section 2. 
 
 140. Every surface of the second order has at least 
 one diametral plane which is perpendicular to the chords 
 bisected by it. 
 
 Let the equation to the surface be 
 ax’ + by? +c2°4+2a'ys+2b'za4+2cvyt+2a vt+2b" y+2c'%+d=0, 
 
 and v=mz, y= nsx, the equations to the line to which a 
 system of chords is parallel; then the equation to the plane 
 which bisects the chords is (Art. 137) 
 
 (am+en+b)xut (bnt+em+a)yt+(c+hm+a'n)s 
 ed Meh +o = 0. 
 
 and our object is to shew that real values can be assigned 
 to mand nm, such that this plane shall be perpendicular to 
 the chords. The conditions for this are (Art. 25) 
 
 am+en+b bn+em+t+a 
 
 ee ee OE wat te 
 , 9 
 c+bm+an c+bm+an’ 
 
 from which, by eliminating one of the unknown quantities 
 
 m or n, we shall obtain a cubic equation which will always 
 
 give a real value for the other; and the direction of the 
 S~ 
 
114 
 
 system of principal chords will be determined. But we shall 
 obtain a more symmetrical result by assuming for the un- 
 known quantity 
 
 c+bm+an=s, ors—-c=bm+an, 
 then am+cen+b'=ms, or m(s—a)=cn+b e+ (1): 
 
 bn+cm+a=ns, orn(s—b)=cm+a 
 Hence, determining m and 7 from the two latter equations, 
 
 m {(s—a)(s—b)-c*} =U (s—b)+ ae 
 
 nm {(s —b) (s—a) —c”} =a'(s —a) + yap 
 
 And substituting for m and m in the former, we have 
 (s —a)(s— b)(s —c)—a*(s —a) —b?(s — b) —c?(s—c) -2a'b'c'=0, 
 or 8 —(a+b+c)s°+ (ab+ ac+be-a’—b’—c")s | 
 — (abe —aa*— bb”? — cc? + 2a'U'c’) = 0. 
 
 This equation, being of an odd degree, will always have 
 one real root which substituted in (2) will give real values 
 for m and»; and therefore in every surface of the second 
 order there is at least one principal plane, or, which is the 
 saine thing, one system of principal chords. Also there can- 
 not be more than three, unless the particular form of the 
 proposed equation of the second order should render any two 
 of the equations (1) identical, in which case m and nm would 
 be indeterminate, and the number of principal planes would 
 be infinite. 
 
 Cor. ‘That the cubic has all its roots real, may be shewn 
 by putting it under the form 
 
 (s—c) }(s—a)(s—b)—c} — $a"(s—a)+b°(s—b) +2a'b'c? =0, 
 
 and substituting for s, a and ;3 the roots of (s—a)(s—b)-—c?=0. 
 The results of these substitutions, since (a ~ a) (a — b) = oF. 
 
 (a = B)(b- B) =e, are 
 we fa'\/a —-axt ba — 6}? and + fa'\/a — B + b'\/b — B}*, 
 
115 
 
 for upon solving the equation (s — a) (s — b) = c”, it is easily 
 seen that one of its roots a is greater than both a and 8, and 
 the other 8 is less, Therefore since +, a, 3, — 0 when 
 substituted for s give results +, —, +, —, there is one root 
 greater than a, another between a and #3, and a third less 
 
 than p. 
 
 141. Every variety of surfaces of the second order referred 
 to rectangular co-ordinates is comprehended, without exception, 
 in the equation 
 
 Av’? + By + Cx? +24e4+2By4+2Cxe+D=0. 
 
 Let the surface be represented by the general equation of 
 the second degree f(a, y, s) =0, and let it be referred to 
 three new rectangular axes Ow’, Oy’, Oz’, by the substitutions 
 of Art. 90, and let the transformed equation be 
 
 Ax’ it By”? i Cz” + 2 A'y’s! ate 2 RB’ =x’ ae 2 C'a'y’ 4 2 A" a 
 +2B"y' +2C"s'+D=0; 
 
 also let the equations to a line parallel to a system of principal 
 chords (the existence of which in every case is certain) referred 
 to the new axes, be a’ = m3’, y’=nz’, then m and n satisfy 
 the conditions 
 
 Am+Cn+ B=m(C+ Bm+ An) 
 Bn+Cm+A=n(C+Bm+ An). 
 
 Suppose now one of the new axes, that of s' for instance, to 
 be parallel to the direction of the principal chords; therefore 
 m=0, n=0; consequently we must have A’=0, B’=0. 
 Hence whenever one of the rectangular axes is parallel to the 
 direction of a system of principal chords, the general equation 
 is freed from two of the rectangles, and takes the form 
 
 Aw? + By? + Cx? +2C'a'y'+2 A"a' +2 B’y' +2C 2’4+D=0...(2). 
 
 This equation comprehends all surfaces of the second order 
 without exception, and it may be still further reduced if, 
 without altering the axis of x’, we turn the axes of a’ and y/ 
 in their own plane through an angle @, so as to make the 
 term involving ay’ disappear, by the substitutions (Art. 94.) 
 8—2 
 
116 
 
 a’ = x, cosg — y, sin ®, 
 y =a, sind + y, cos d; 
 this gives — 2cos@ sin (4 — B) + 2C’ Scos* ph - sin? hb} = 0, 
 2C" 
  tan2@ =——_., 
 or tan2@ Gap 
 
 which value, being real and always admissible even when 
 A = B, shews that equation (2) may be reduced to the form 
 (suppressing accents) 
 
 Aa’ + By + C+ 2Ae+2By+2C2+ D=0; 
 
 which (the co-ordinates being rectangular) comprehends all 
 surfaces of the second order. 
 
 142. It is at this point of the reduction of the general 
 equation of the second order, that the separation of the sur- 
 faces represented by it into two classes takes place. 
 
 1. If none of the squares of the variables are wanting 
 in the equation 
 
 Aa’? + By +024+2Ae+2By+2C2e+D=0, 
 
 i.e. if A, B, C are all different from zero, by writing 
 eth, y+k, =+1 for x, y, x and determining h, k, 1 by 
 the conditions 
 
 Ah+A=0, Bk+B=0, Cl+C'=0, 
 
 (which will give finite values for h, k, 1) the origin will be 
 changed, but the directions of the axes unaltered, and the 
 equation will be reduced to the form 
 
 Av’ + By? + Cz? =D... (1) 
 which represents the first class of surfaces of the second order. 
 2. If one only of the squares is wanting, A for instance, 
 and the corresponding coefficient A’ is different from zero, 
 
 then the term involving w cannot be made to disappear, for 
 that would give / infinite; but instead of that we may deter- 
 
117 
 
 ‘mine f# so as to exterminate the constant term, and determiné 
 k and 7 by the same conditions as before, and the equation 
 will be reduced to the form 
 
 Bye OS = 2A (2) 
 
 which represents the second class of surfaces of the second 
 order, ‘The figures and properties of these two classes of 
 surfaces have already been discussed, (Art. 53). 
 
 143. There are still two particular cases to be examined, 
 belonging, as we shall see, to the two sorts of cylindrical 
 surfaces of the second order. 
 
 First, suppose that the coefficient of one of the squares, 
 A for instance, vanishes, and at the same time the correspond- 
 ing coefficient A’=0, without either of the other squares dis- 
 appearing; the equation is then 
 
 By’ + C2*4+2By4+2C2+D=0, 
 
 which, containing only two of the variables, represents a 
 cylinder perpendicular to the plane of yz on an elliptic or 
 hyperbolic base; and if the center of this curve be taken for 
 the origin, the equation will be reduced to the form 
 
 By iG ei 
 
 which is deducible from equation (1) by making 4 = 0. 
 
 Secondly, suppose only one of the squares to remain in 
 the equation, x? for instance; then writing »+/ and a+h 
 for s and #, and determining # and 7 so as to exterminate 
 the term containing x, and the constant term, the equation 
 becomes 
 
 C#+2Aa7+2B y=0; 
 
 now, without altering z, turn the axes of w and y in their 
 own plane through an angle @ such that 4’ sin p = B' cos , 
 
 then the equation will be reduced to the form 
 
118 
 
 which represents a cylinder perpendicular to the plane of wy 
 on a parabolic base, and is deducible from equation (2) by 
 putting B= 0. 
 
 It is unnecessary to examine the case where the three squares 
 disappear, as no equation of the second order could ever by 
 transformation of co-ordinates be reduced to that form; the 
 cases in which it represents a cone, or two parallel planes, 
 have been noticed. 
 
 144. Hence, we conclude, that all surfaces of the second 
 order with all their varieties, are comprised in the two classes 
 represented by the equations 
 
 Ax’? + By + C2 =D, 
 By? + C2? =2Aa, 
 
 the co-ordinates being rectangular. In the first class, each 
 of the co-ordinate planes is a principal plane; therefore there 
 are three systems of principal chords, and consequently the 
 three roots of the cubic in (Art. 140) are all real. In the 
 second class, only the planes of s# and wy are principal 
 planes; therefore there are at least two systems of principal 
 chords; the third system, determined with the others by the 
 cubic equation just referred to, must therefore be also real, 
 but the corresponding principal plane is situated at an infinite 
 distance. Hence it results from this discussion, as well as 
 from the form of the equation (as proved in Cor. Art. 140), 
 that the three roots of the cubic equation which determines 
 the positions of the principal chords are always real. 
 
 145. By means of the preceding results, we are able to 
 determine the species and form of the surface represented by 
 any proposed equation of the second order, without recurring 
 to the laborious process of transformation of co-ordinates. 
 
 We must first ascertain whether the surface has a center 
 by the method of Art. 134, and if it has, find the co-ordinates 
 of the center h, hk, 1 from the equations which result, by 
 equating to nothing the three derived equations of the first 
 
 ——— 
 
119 
 
 order, w, y, # being replaced by h, k, 1; then taking that 
 point as origin, the equation is reduced to 
 
 aa’ + by? +c3x?4+2ayx4+2bsa+26ay4+ah+b"k+c'l+d=0, 
 
 from which the species of the surface and the values of the 
 real or imaginary axes can be readily determined, as will be 
 seen in the next Art. In the succeeding Articles, the charac- 
 teristics of the different sorts of surfaces that have not a center 
 
 will be deduced. 
 146. The equation 
 au’ + by? + cz? + 2a ys + 2b ea + 2uvy =d, 
 
 (d being a positive quantity) belongs to an ellipsoid, hyper- 
 boloid of one sheet, or hyperboloid of two sheets, according 
 as the cubic equation 
 
 (s —a)(s—b) (s —c) —a?(s — a) —b?(s—b)-c?(s—c) -20'b'c'=0, 
 gives for s three positive values, two, or one. 
 
 Let w=mz, y=nsz, be the equations to a line through 
 the origin, which will be a diameter of the surface, since 
 the origin is the center, and let the length of the diameter 
 be denoted by 27; then the chords parallel to it will be 
 bisected by a plane, of which the equation is 
 
 am+cen+b)a+(bn+emt+ta)y+(c+dm+an)s=0; 
 Yy 
 
 and if this plane be perpendicular to the chords, or if 27 
 be a principal diameter, 
 
 making c+ bm+an=s, 
 
 am+en+b=ms 
 we must have }..0) 
 
 bn+cecm+a'=ns, 
 from which eliminating m and 7, as in Art. 140, we find 
 
 (s—a)(s —b)(s—c) — a? (s— a) —b?(s—b)—c? (s—c) -2a'b'c'=0; 
 
120 
 
 also multiplying the two latter equations by m and m respec- 
 tively, and adding them to the former, we find 
 
 s(1 +m? + n*) = am? + bn? 404 2a'n + 2b'm + 2Cmn. 
 But if w, y, x be the co-ordinates of the extremity of the 
 diameter, 
 
 9 
 ~ 
 
 ra eP+ypPrs= (1+ m +n’) 2; 
 also since wv, y, % are co-ordinates of a pcint in the surface, 
 3 (am? + bn? +c0+2an+2bm+2cemn) =d; 
 
 d d 
 
 2? Pim 
 Ss 
 
 tw) 
 
 ae y = or r = 
 
 3° x 
 —.sl+m+n')= 
 s 
 
 ee 
 
 Ss 
 
 Now d is a positive quantity, therefore according as s has 
 three positive values, two, or one, r has three real values, 
 two or one, or the surface has three real principal diameters, 
 two, or one, and is therefore an ellipsoid, hyperboloid of one 
 sheet, or hyperboloid of two sheets. 
 
 147. Hence, when we have an equation of the second 
 order reckoned from the center, and the constant term forms 
 one member of the equation and is positive, if with the 
 numerical coefficients we form the above cubic, and observe 
 the signs of its terms, we shall determine the nature of the 
 surface represented by the proposed equation. For since all 
 the roots of the cubic are real, there will be three positive 
 roots, two, or one, according as there are in it supposed com- 
 plete, three changes of signs, two, or one. If the cubic offers 
 no change of signs, but only continuations, its three roots are 
 all negative, and the three axes of the surface are imaginary, 
 and consequently the surface itself imaginary. 
 
 Cor. If we solve the cubic, we may then easily calculate 
 the lengths of the axes, and fix the position of each by means 
 of the values of m and n, corresponding to each value of s, 
 given by equations (1). 
 
 Ex. 1. 2a°+5y°+32°42yx%-420-20y+2u+8y-624+8 =0, 
 
121 
 
 Forming the three derived equations and replacing a, y, z 
 by h, k, J the co-ordinates of the center, we get 
 
 Diy ra FEA a Te Bo as: 
 10k +21 —2h+8=0,> which give = —1, 
 614+2k-—4h—6=0, b= 2. 
 
 Then the equation, referred to the center, and the cubic for 
 determining the axes, become respectively 
 
 20° + by? + 32° + Qyz—430 —-Qwy=—-14+446-8=], 
 s§—10s°+ 78+ 388 =0; 
 
 and as this complete equation has all its roots real and has 
 two changes and one continuation of sign, it has two positive 
 roots and one negative root, and the surface is a hyperboloid 
 of one sheet. 
 
 Ex. 2. v0 +2y? 4+ 3274+ 2yx4+22u+4+2uyt+ut+y+s=1. 
 The co-ordinates of the center will be found to be 
 h= — i, b=0, 1=.0; 
 and the equation referred to the center becomes 
 a+ 2y + 32° 4+ 2ye + 2Q2u + 2aHy =F. 
 Hence the cubic becomes 
 8 —6s?+8s—2=0, 
 
 having three real roots; and as it is.complete and has three 
 changes of sign, it has three positive roots, and therefore 
 represents an ellipsoid. 
 
 Ex.3. ystxuv+ay=a’. 
 
 Here the cubic is s* — 3s — 2 =0, which has one positive 
 root and cannot have more; therefore the surface is a hyper- 
 boloid of two sheets. | 
 
 Ex. 4. «4+ 2y°+3ys-2ey-6x+7y+624+7=0. 
 
122 
 
 The co-ordinates of the center will be found to be 
 h=1, k=—2, t=, 
 and the equation referred to that point 
 uv +2y°? + 3y% —2xy =0, 
 
 wanting the constant term; and as it is satisfied by w =0, 
 y =0, it contains the axis of x, and represents a cone and 
 not a point. 
 
 148. When the proposed equation 
 axv+by+ex 42a yxet+2b su42Cay+2a vVt+2b y+2c' zx +d=0, 
 is such that 
 abe — aa® — bb”? — cc” + 2ab'c = 0, 
 
 the surface represented either has not a center, or it has a 
 central line or plane (Art. 135), and admits of four varieties, 
 viz., paraboloids, parabolic cylinders, elliptic or hyperbolic 
 cylinders, and a system of two parallel planes; for each one 
 of which, certain conditions must be satisfied which are not 
 all true for the others. Hence we can determine the dis- 
 tinctive characteristics of each of these surfaces. 
 
 1. For paraboloids, since the co-ordinate planes cannot 
 be all parallel to the axis of the surface, and since it is 
 only such planes which intersect the surface in a parabola 
 (Art. 103), one of the quantities 
 
 ab, (1) 
 
 (which determine the nature of the traces on the co-ordinate 
 planes) must be different from zero, and negative in the case 
 of the elliptic, and positive in that of the hyperbolic para- 
 boloid ; also one at least of the co-ordinates of the center will 
 be infinite, and therefore one of the equations for determining 
 those co-ordinates imaginary. 
 
 a*—be, b?-—ac, c 
 
 2. For parabolic cylinders, since all sections are either 
 straight lines or parabolas, each of the quantities (1) vanishes, 
 
123 
 
 and one of the equations for determining the co-ordinates of 
 the center will be imaginary. 
 
 8. For elliptic or hyperbolic cylinders, the equations for 
 determining the co-ordinates of the center are reduced to two 
 distinct equations, and one at least of the quantities (1) is 
 different from zero, and is negative in the former, and positive 
 in the latter case. Also, by taking sections parallel to the 
 co-ordinate planes, it will be necessary further to examine 
 whether the cylinder is wholly imaginary, or reduced to a line, 
 or to a system of two planes not parallel. 
 
 4. For a system of two parallel planes, the equations for 
 determining the co-ordinates of the center must be reduced to 
 a single distinct equation, and the three quantities (1) must 
 each vanish. It will be further necessary, by taking sections, 
 to examine whether the two planes are confounded in one, or 
 are imaginary. 
 
 Ex.1. a’ -—2y’?-— 3y2 4+ 32a + ay +42 =0. 
 
 The equations for determining the co-ordinates of the 
 center are 
 
 20+3%3+Yy=0, 
 —4y—-32+4+07=0, 
 —3y+ 3%+4=0; 
 
 and since the last subtracted from the sum of the two former 
 gives 4=0, this impossible equation shews that the surface 
 has not a center. Also, c? — ab = (4) + 2, which is different 
 from zero and positive; therefore the surface is a hyperbolic 
 
 paraboloid. 
 
 Ex.2. a? +y? +92" + 6y2 — 62a —2ay + 2u—4z% =0, 
 The three derived equations are ] 
 
 2e —-62—-2y+2=0, 
 
 2Qy + 6% —2e@ = 0, 
 
 18x + 6y — 6% —4=0, 
 
124 
 
 the two former of which give 2=0, therefore the surface has 
 no center; and the traces on the co-ordinate planes cannot be 
 ellipses or hyperbolas, therefore the surface is a parabolic 
 cylinder. 
 Ex.3. #® -y-24+2yx+e+y—-x =a. 
 The derived equations are 
 2%+1 =u), 
 —-2y+22+1=0, 
 which are reduced to two distinct equations; also the traces 
 on wy and ws are hyperbolas; therefore the surface is a 
 hyperbolic cylinder; except when a@=0, when it represents 
 two planes intersecting in the axis of the cylinder whose 
 
 equations are 2v+1=0, 2x -—2y+1=0; for when a=0, 
 the equation is resolvable into (w + y — 2) (w@-y+2+41)=0. 
 
 Similarly, wv + 3y? + 42° -—6yz —2%x” =a represents an 
 elliptic cylinder; except when a =0, when it represents the 
 straight line « = %, y = #, forming the axis of the cylinder; 
 for it may then be written (@ —%)? + 3(y—x)?=0. If a be 
 negative, the surface is imaginary. 
 
 Ex. 4. w+ 4y°+ 2 4+4y2-22a -40y43au—-6y—32=0. 
 
 The derived equations are 
 
 20-23 —-4y+3=0, 
 8y +42 —-4v—-6=0, 
 23+ 4y —-24%-—-3=0, 
 which are equivalent to only one equation; therefore, the 
 
 proposed equation can only represent two parallel planes; and 
 putting x = 0, the trace on wy has for its equation 
 
 w+ 4y? —4vy + 3x —-6y = 0, or (w — 2y)’? +3 (x — 2y) = 0, 
 
 representing two real straight lines; consequently the proposed 
 equation represents two separate parallel planes, and, as we 
 see, it may be written (v7 — 2y — x) (vw- 2y—%+4 3) =0. 
 
125 
 
 149. To find the relations among the coefficients of the 
 general equation of the second order when it represents a 
 surface of revolution. 
 
 Let av? + by? + c2°+2a yx 4+2b'ax 4+ 2c ary + 2a ex 
 
 +2b°y4+2c'x+d=0, (1) 
 
 be the general equation to the surface; and let a, 8, y be the 
 co-ordinates of a point in the axis of revolution. “Then a 
 
 sphere whose equation is 
 (v7 —a)’+(y—-B)y+ @-y)- R=0, (2) 
 will intersect the surface in two parallel planes 
 i pircaih. eit 
 len +my+nz =p. 
 
 Now if \ be an arbitrary multiplier, 
 (1) +A (2) =0 
 
 gives the intersections of the surface with any sphere such as 
 that above mentioned ; hence this equation must be identical 
 
 with 
 O=(la+my+nz—p)(le+my+nzx—p’); 
 
 .O0O=PH +m yt ns? +2mnyx4+ 2lnae + 2lmay + &e. 
 is identical with 
 
 O=(a+A)a’?+(b+A)y?+ (C+A) +420 YX 42h az42Cry+&e.; 
 
 -P=at+aA, mM=b+r, W=c+”, 
 mn =a’, In =0’, lm=c 
 b'e! ; ae’ ih 
 pel=sat+nr, —-=m=b+)d, —-=Wect+d; 
 a b C 
 bo! a’e a’b’ 
 ° ee eer b = —— —"¢, 
 a b 
 
 are the required conditions. To this must be added that each 
 of these equated quantities, being equal to — A, must be finite. 
 
126 
 
 Cor. Hence the plane through the origin to which the 
 circular sections are parallel has for its equation 
 
 Veutacy+abz =0, 
 consequently all chords of the surface parallel to this plane are 
 principal chords. 
 
 Also if we substitute for a, b, c the values furnished by 
 these equations in the cubic (Art. 140), it will take the form 
 bic, 20 0 we ean 
 
 (+r {srn-“ ~~ hao, 
 
 Hence, when the above conditions are satisfied, the cubic 
 for determining the directions of the principal chords has two 
 roots equal to — A, so that of the three systems of principal 
 chords two may be drawn in any manner parallel to the plane 
 represented by the above equation. Moreover when the sur- 
 face has a center the equal values of s lead immediately to 
 the value of its equatoreal axis (Art. 146), and the remaining 
 value of s to that of the polar axis, 
 
 150. To find the equations to the axis of revolution. 
 By comparing the terms of the first order in the two 
 
 identical equations of the preceding article we get 
 2a"—-2r\a=—-I(pt+p’), 2b"°-2\B =—m(p+p), 
 2c" —- 2rAy=—n(p+p'); 
 but mn=a’, In=b*, Im=c’, (3) 
 *. a (a —da) = 5 (b" -AP) =e (c" —Ay); 
 or, replacing a, 3, y by a, y, s, and X by its values, the 
 relations amongst the co-ordinates of the center of the sphere 
 
 that always intersects the given surface of the second order in 
 two parallel planes, become 
 
 i A a’a”’ u( : bp”! ( ant 
 TE Ls emerges garry Pi SB re rrr mh eae recs 
 nae be Leh DN & +o ay) 
 
 which represent a straight line forming the axis of revolution 
 of the surface; and, as appears from equations (3), perpen- 
 dicular to the parallel planes. Although the results of this 
 
127 
 
 and the preceding Article may be so modified as to embrace 
 all particular cases, yet when a proposed equation wants 
 several terms, it will generally be better to apply the direct 
 investigation to discover whether or no it represents a surface 
 of revolution. 
 
 151. Three diametral planes are said to be conjugate 
 to one another, when each bisects the chords which are 
 parallel to the intersection of the two others; and the inter- 
 sections in that case are called conjugate diameters. 
 
 When a surface admits of three planes of this sort and 
 they are taken for the co-ordinate planes, its equation supposed 
 algebraic can only contain even powers of the three variables 
 a, y, x Hence the principal planes of a surface of the second 
 order having a center, are conjugate to one another, since 
 the equation referred to them is 
 
 Av? + By? +C2? =D; 
 
 and this, as we have seen, is the only rectangular system of 
 co-ordinate planes which can give the equation of this form, or 
 which can be conjugate to one another. 
 
 152. We shall now shew that there is an infinite number 
 of systems of diametral planes oblique to one another which, 
 taken for the co-ordinate planes, will make the general equation 
 of the second order to be of the above form, that is, free from 
 the terms involving za, wy, yx, and which are therefore con- 
 jugate to one another, 
 
 Let w=mz, y=nz be the equations to any diameter 
 Ox’ (fig. 50), then the equation to the diametral plane «’ Oy’ 
 conjugate to Oz’, is (Art. 137), 
 
 df (x, Y, 2) ii AAC Y, &) = df (x, Ys %) as 
 x 
 
 ae O 
 iy: dy dx ; 
 
 or, mAv+nBy+Cz=0. 
 Since the diameter O2’ is parallel to the chords which 2 Oy’ 
 bisects it must according to the definition be the intersection of 
 
 the two planes which are conjugate to wv Oy’; and those planes 
 will cut the plane #’Oy’ in two lines Oa’, Oy’, such that if 
 
128 
 
 together with Ox’ they be taken for a system of oblique axes, 
 the equation to the surface will only contain even powers of 
 the three variables, and therefore be of the form 
 
 Ae? + Byy? + Ciz" = D,. 
 
 But if we make x’ = 0, the result A,a”’ + B,y? = D, is the 
 equation to the section a Oy’, and by its form shews that 
 the axes Ow’, Oy’ are.conjugate diameters of that section ; 
 hence it follows, that any diameter Ox’ being proposed, if 
 in the section made by the diametral plane which is conjugate 
 to it we draw any two conjugate diameters Oa’, Oy’, at 
 pleasure, we shall determine three planes z’Oa’, x’ Oy’, y' Ox’ 
 which are conjugate to one another; the number of such 
 systems is therefore unlimited ; and the equation to the surface 
 when referred to a system of conjugate diameters as co-ordinate 
 axes, may be put under the form 
 
 a’, b’, c’, being the distances, real or imaginary, from the center 
 at which the surface cuts them. 
 
 Cor. If Oz’ be in a principal plane, the tangent plane 
 at the extremity of Ox’ and consequently the diametral plane 
 conjugate to Ox’ will be perpendicular to that plane, but not 
 perpendicular to Oz’. Also if Oz’, Oy’, be axes of the section 
 w’ Oy’, they will be at right angles to one another; or if Oz" 
 coincide with an axis of the surface, it will be perpendicular 
 both to Ow’ and Oy which will lie in a principal plane; but 
 in no case will three conjugate diameters be mutually at right 
 angles, unless they coincide with the axes of the surface. 
 
 153. We have seen (Art. 138) that any diametral plane 
 is parallel to the tangent plane, applied at the extremity 
 of the diameter to which it is conjugate. Hence, in a system 
 of conjugate diameters, the tangent plane at the extremity 
 of each is parallel to the plane of the two others. Also, 
 if through the extremities of each of a system of conjugate 
 diameters we draw planes parallel to the plane of the two 
 others, we shall form a parallelopiped, which is said to be 
 
129 
 
 constructed upon the diameters, and which in the case of the 
 ellipsoid will be circumscribed about the surface. We shall 
 now shew that the known properties of conjugate diameters of 
 curves of the second order may be extended to surfaces of the 
 second order. 
 
 154. Ina surface of the second order, that has a center, 
 the sum of the squares of any system.of conjugate diameters 
 is equal to the sum of the squares of the axes; also the 
 volume, and the sum of the squares of the faces, of the 
 parallelopiped constructed on any system of conjugate dia- 
 meters, are respectively equal to the volume, and the sum of 
 the squares of the faces, of the rectangular parallelopiped 
 constructed on the axes. 
 
 2 y 2 , 
 
 Let wtiatat 1 be the equation to a surface of 
 the second order, referred to a system of conjugate diameters, 
 the inclinations of which are Zyz=A, Z2v=p, Ley =r; 
 and let r=mz, y=nz, be the equations to any diameter ; 
 then the equation to the diametral plane conjugate to this 
 diameter is 
 
 me ny z 
 TE DER LES Fae topo de a ae 
 a b c 
 
 and if it be perpendicular to the diameter, we have (Art. 86) 
 mec*(1 + mcosu + ncosr) = a?(m + CoS v + COS 1), 
 ne*(1 + mcosu +n cos)) = 6" (nm + m cos v + cos dX). 
 Let r =c"(1+mcosp + 2COSA) 3; 
 
 “mr = a" (m +n Cos vp + COS p), 
 
 nr° = b?(m + mcos v + 0s A) 3 
 
 Les 10. va Ne 
 id Sate & +—+ =] =1+m’+n?+2mncosyv+2mMcosu+2N Cosas; 
 
 1 
 a’ 
 
 *. (as in Art. 146) ris the length of the semidiameter whose 
 a 
 
130 
 
 equations are a= mz, y=nzx. Also if we determine m and 
 nm from the two latter equations, and substitute them in the 
 former, the result is 
 
 (7° —a’*) (7? —b”) (r? —c’*) —a’*b? (1? —c”) (cosy)? —a’*c?(r* —b”) (cosu)? 
 — be? (r* — a”) (cos d)* — 2ab”c? cos p cos vy cosA = 0, 
 or 7° — 7" (a? +b" + 6") 47°f (a'U'siny)’?+(a'c'sinu)?+(b'c'sind)”} 
 — (a'b’c’)*$1 + 2cosd cos u cosy — (cosd)* — (cos)? — (cos v)*} =0, 
 
 But if a, b, c be the three semiaxes of the surface, then 
 a*, b*, c are the values of r? in the above equation; there- 
 fore, by the theory of equations, we have 
 
 a? +b7° 4+ Pra V+ +0’, 
 (a’b’ sin v)?+ (ac sinu)?+ (0'e' sind)? = (ab)? + (ac) +(bc)*.... (2); 
 (a’b’c’)* $1 + 2 cos X Cos pm Cos v 
 — (cos \)? — (cos nx)” — (cos v)*} = (abc)’.....(3), 
 
 which are the three required results, the first member of 
 the third being the well-known expression for the square of 
 the volume of a parallelopiped in terms of its three edges, 
 and their inclinations to one another. 
 
 Cor. When one or two of the axes are imaginary, 
 there will be an equal number of the conjugate diameters 
 imaginary; and it will be necessary to change the signs of 
 the squares of these axes and these diameters in the above 
 equations. 
 
 155. The preceding results may also readily be arrived 
 at by the following method. 
 
 Let Ow, Oy, Ox (fig. 29) be the semiaxes of the sur- 
 face a, b, c; Ow’, Oy’, Oz’ any system of semiconjugate 
 diameters a’, 6’, c’; let the plane of a’y’ intersect that of 
 avy in the 4 diameter Ow,=a,; also let Oy,=b, be the 
 
 semidiameter of the curve #,x’, which is conjugate to Oa; 
 
 a+b, =a? +b” 
 
131 
 
 Now the plane aw Oy' is by supposition conjugate to 0, 
 therefore Ox’, Oy,, Ox,, form a system of conjugate diameters, 
 or Ow, is conjugate to the plane x’ Oy. 
 
 Let the plane x’ Oy, intersect wy in the semidiameter 
 Oy, = b,, then this plane must contain Ox; for being con- 
 jugate to Ow, in a principal plane, it must be perpendicular 
 to that plane (Cor. Art. 152); hence Ox,, Oy,, Ox form a 
 system of semiconjugate diameters, (for Oz, Oy, are the 4 axes 
 of the section y,y,%) and any two are semiconjugate diameters 
 of the plane section in which they are situated ; 
 
 U0; Ho =0, ey 
 a+b =a + bi; 
 therefore, adding these equations to the former, 
 a’ “fs b? += q? ib? 6. 
 
 Again, employing the same auxiliary systems of diameters 
 as above, and denoting each parallelopiped by its three edges, 
 we have 
 
 vol. (a’, b’, c’) = vol. (a, be, ¢’), 
 
 for these figures have the same altitude, viz. the perpendi- 
 cular from ’ on the plane of «’y’, and equal bases, viz. the 
 parallelograms constructed on the two systems of 4 conjugate 
 diameters a’, 6’, a,, b,, belonging to the same curve in the 
 
 plane of ay’. Similarly, 
 | vol. (a, b,, c’) = vol. (a), 0, ¢), 
 vol. (a), bj, c) = vol. (a, 6, c) ; 
 .. vol. (a, b’, c) = vol. (a, 8, c). 
 
 Cor. If the conjugate diameters 2a’, 2b’, 2c’ are each 
 equal to 2R, then 3R*? =a’? +b’?+c’; also since there are 
 only two equations, viz. (2) and (3) Art. 154, to determine 
 the angles of inclination A, uw, v of the conjugate diameters, 
 there may be an infinite number of systems of equal conjugate 
 diameters, their extremities all lying in the intersection of the 
 surface and a concentric sphere radius = R. 
 
 9—2 
 
132 
 
 And of all systems of conjugate diameters of an ellipsoid 
 the principal diameters have their sum a minimum and the 
 equal diameters their sum a maximum. For in any proposed 
 system, if there were two not perpendicular to one another, 
 by substituting for them the axes of the section of which they 
 are diameters, we should obtain a system whose sum is less 
 than that of the proposed system. Again, if there were two 
 not equal to one another, by substituting for them the equal 
 diameters of the section to which they belong, we should 
 determine a system whose sum is greater than that of the 
 proposed one. And in this manner we could shew that no 
 oblique system could have their sum a minimum, and no 
 unequal system their sum a maximum. 
 
 156. In surfaces of the second order not having a center, 
 represented by the equation to rectangular co-ordinates 
 
 By + Cz2*°=2/4' a, 
 
 the planes of xa and vy only are diametral and principal, 
 each bisecting perpendicularly the chords parallel to the 
 intersection of the other with the plane of yz, which is the 
 tangent plane at the vertex. In this case no three diametral 
 planes can be conjugate to one another; for, taking a system 
 of chords parallel to a line whose equations are v = mz, y = 72; 
 the equation to the diametral plane conjugate to them is 
 
 Therefore all the diametral planes are parallel to the axis 
 of a, and consequently their intersections, which are the 
 diameters of the surface, are parallel to that axis, and cannot 
 therefore form a system of co-ordinate axes. 
 
 But we can find an infinite number of oblique co- 
 ordinate planes related to one another in the same manner 
 as the three rectangular planes mentioned above are, viz. so 
 that two, ’# and ay’, shall be diametral planes, and each 
 of them conjugate to the chords parallel to the intersection 
 of the other with y's’; the latter being a tangent plane to 
 the surface at the extremity of the diameter in which 3’ a’ 
 
133 
 
 and a’y’ intersect; and the equation to the surface when 
 referred to them will therefore preserve the same form. 
 
 For let a’ O'y’ (fig. 51) be the diametral plane represented 
 by equation (1), and which, being parallel to the axis of the 
 surface, cuts it in a parabola AO’B. In AO'B take any point 
 O’, and draw O’’ parallel to the chords to which the diametral 
 plane is conjugate. Then the two co-ordinate planes which 
 are to go along with this diametral plane will pass through 
 O's’, and cut the diametral plane in two lines O'v’, O’y’, such 
 that being taken for the co-ordinate axes they shall give the 
 equation to the surface under the form 
 
 Byy® + Cis- = SA it. 
 Therefore, making ’ = 0, the equation to AO’B is 
 By” = 2 Aa’. 
 
 which by its form shews that it is referred to a diameter 
 of the parabola and a tangent to the parabola at the extremity 
 of that diameter. Hence the position of the second diametral 
 plane x’ O'x’, and also of the third co-ordinate plane x’ O'y’, 
 is determined, which latter, since it passes through two tangent 
 lines to the surface O'y’ and O'’, is a tangent plane to the 
 surface at O'; and as not only the plane 4O’B is arbitrary, 
 but also the position of the point O’, the number of systems of 
 oblique co-ordinate planes similar to the above is unlimited. 
 
 157. When the general equation of the second order 
 represents a paraboloid, to find the position and magnitude of 
 the surface. 
 
 Let aa’ + by? + c2x* + 2a'yxz4 2b ax4+2Cuy+2a' a 
 
 + 2b’y4+2c's+d=0, 
 be the equation, then 
 D = aa’ + bb”? + cc? — abe — 2a'b'e' =0; 
 
 consequently the cubic which determines the directions of the 
 principal chords has one root s=0, and the corresponding 
 values of m and 7 are (Art. 140), 
 
 m(ab—c”) =a'c — bb, n(ab—c”*)=)c'- ad’. 
 
134 
 
 Since the plane bisecting the chords for which s =0 is at an 
 infinite distance, these chords are in the direction of the axis 
 of the paraboloid; and therefore the equations to a line 
 through the origin parallel to this axis are 
 
 A Y Z 
 MLS PA ea a OL) 
 ac—-bb be-—ad ab-c® 
 Also if w,y,% denote the co-ordinates of the vertex, 
 expressing that the normal at that point must be parallel 
 to (1) we get the equations 
 
 1 du 1 du 1 du 
 ab—c? dz ac’—bb' dz wc'-—aa’ dy’ 
 which, together with the equation to the surface wu = 0, will 
 determine a, y, x the co-ordinates of the vertex. 
 
 Next to find the values of the parameters of the principal 
 sections, let 
 
 ’ , M 
 aa + by? + cx? + 2ays + 2az + 2Cuy + Dies 
 where M=a"N+b’N'+c"N"4+dD (Art. 134), be the 
 
 equation to a surface of the second order referred to its center ; 
 then the roots of the equation 
 
 D D? 4 
 s+(a+b+0c) By peat as Lae teat ad a oe) ps _- 0, 
 PD D? D' 
 “or gf ln Feit ae - ays 7 0 (Att 146) 
 
 are the squares of the reciprocals of the semi-axes a, 3, +. 
 Now suppose D very small, then one value of s will be very 
 
 1 1 1 
 small, let this be =; and the remaining values of s, ead 
 a p yy” 
 will be nearly equal to the roots of 
 PD QD 
 
 s 4+ ——'s + ——_ = 0; 
 uM** MM? 
 
 aye 1 PD\? Pet 
 eal er ae I 
 
135 
 
 D' 1 QD? 
 (aByy MP’ (Bye MP’ 
 we war tt Q Q’ ak \ tea. C)* 
 [oe eed ~ 
 
 Consequently, the squares of the reciprocals of the 4 para- 
 
 meters of the ue ag sections of the paraboloid, being the 
 
 ] 2 Q . 
 and’ @* = Dr” since 
 
 values af poh — 
 
 eo 4.9 
 
 p= ("= 2Q) =. p+ 
 
 when D = 0, are the roots of the quadratic, 
 Q' 
 im 
 
 If 1,7, be those 4 parameters, we have 
 
 1 ut (Jay Jakes Cle 
 pat Cpe ogy! 2S 
 pt pn ' Var Wr aw 
 et We Es - 
 sa ll Q ? 
 
 which shews that paraboloids represented by the general 
 equation of the second order, have the parameters of their 
 principal sections in the same ratio, or are similar to one 
 
 2 
 
 another, when has the same value. Also the surface will 
 
 be an elliptic or hyperbolic paraboloid according as Q is posi- 
 tive or negative; for as a is a real semi-axis and therefore a’ 
 positive, Q and M must have the same sign in order that any 
 surface may exist. 
 
 Cor. Similarly we may find the conditions that two 
 central surfaces of the second order may have their axes 
 proportional, or be similar to one. another. 
 
 As above, the cubic equation 
 
 . 1 e e e e 
 will have roots —, Let this equation in which we 
 a 
 
 1 1 
 
136 
 
 may suppose M = 1 without altering the relations of its roots 
 to one another, be transformed into another whose roots are 
 
 1 it ; 
 
 (Seen tee) male Be 
 
 4 \B Wal ko” Wa pede 
 the result will be found to be 
 
 PQ jie aig 8S es ) 
 Ey) | Slap +3)@-—|——4+1] =0. 
 ale a) 16\D? D* D* as 
 . ae ah eee 
 Now the values of ¢ depend only on the ratios =, — and —; 
 SRO fi 8 
 
 in 
 consequently these ratios will be the same, provided > and 
 
 el 5 hy ae 
 es =p have the same values; these therefore are the con- 
 ditions for central surfaces, when represented by the general 
 equation of the second order, having the ratios of their axes 
 the same, or being similar to one another. 
 
SECTION VII. 
 
 ON CYLINDRICAL, CONICAL, AND CONOIDAL SURFACES, AND 
 ON SURFACES OF REVOLUTION. 
 
 158. Brrore proceeding with the proper subject of this 
 section it may be useful first to introduce the following 
 remarks relative to the number of constants in the equations 
 to surfaces, and to the intersection of surfaces. 
 
 If the general equations of the n™, (m — 1)" &c. orders 
 between two variables x and y be formed, and multiplied 
 respectively by the quantities 1, z, 2’, &c. %", we shall by 
 adding all such products together form the general equation of 
 the 2" order between three variables a, y, x; and the number 
 of constants which this equation will contain will be 
 
 E(m+1)(n+2)+4n(n+1) +h (m—-1)n+...4+$1.2 
 =1(m +1) (n+ 2) (m + 3). 
 
 Consequently, diminishing this by unity we shall have the 
 number of independent constants in the general equation of 
 the 7" order between three variables expressed by 
 
 N=1n(n?+6n+ 11); 
 
 which is also the number of points through which a surface 
 of the nm order can be made to pass, for the co-ordinates 
 of every one of these points being substituted in the equation 
 to the surface would furnish N linear equations for the 
 determination of the constants. 
 
 Let w=0, v =0, be the equations to two surfaces that 
 intersect; then w+XAv=0, where XA is an indeterminate 
 constant different from zero, represents a surface passing 
 through their curve of intersection, since it is satisfied by 
 all values of wv, y, and x that simultaneously satisfy the 
 equations w= 0, v = 0. 
 
138 
 
 Also if w= 0, v =0, be the equations to two surfaces of 
 the mn order passing through N—1 given points, then 
 uw+2Xv=0, since it contains an additional constant ), in- 
 cludes all the surfaces of the m' order that can pass through 
 the given points, which shews that all those surfaces will pass 
 through the intersection of any two of them; therefore all 
 surfaces of the 2“ order that pass through NV — 1 given points, 
 have a common curve of intersection. 
 
 And if a surface of the m™ order be made to pass through 
 a number of fixed points two less than the number sufficient to 
 completely determine it, the surface will also pass through an 
 additional number of fixed points such that added to the 
 former it makes up m* the entire number of points in which 
 three surfaces of the n“ order can intersect one another. 
 
 Since an infinite number of surfaces of the m™ order can 
 be described through the N —2 given points, consider any 
 three of them whose equations are wu =0, v=0, w =03 then 
 the equation w~+Av+uw=O0 (1) where A and pw are in- 
 determinate constants, will include all the surfaces of the 
 nm order that can pass through the given points, because the 
 equation of every such surface could involve only two un- 
 determined constants. But equation (1) will be satisfied by 
 every system of values of x, y, and x that simultaneously satisfy 
 u=0,v=0,w=0; that is, all the surfaces will pass through 
 the points of intersection of any three of them; therefore 
 all the points of intersection must be fixed points, and the 
 N —2 given points will determine the n?— N + 2 remaining 
 fixed points of intersection. | 
 
 Again, if the system of equations u,=0, v,=0 taken 
 together represent the curve of intersection of the surfaces 
 U, =0, v,=0; and w,=0, v,=0 taken together represent. 
 the curve of intersection of the surfaces of w,=0, v,=0; then 
 will 2, + AVv,v; = 0 represent another surface which passes 
 through both these curves of intersection, and is of the order 
 indicated by the greater of the quantities ” or r +s. 
 
 Hence if v=0, v =0, be the equations to two planes 
 intersecting a surface w=0; then «+)vv'=0 is the equa- 
 
139 
 
 tion to a surface passing through the two plane curves of 
 intersection; and if the planes become coincident, w+? = 0 
 represents a surface touching w= 0 along the curve in which 
 the plane v = 0 cuts it. 
 
 A surface of the m‘" order may in general have n (n — ii 
 tangent planes applied to it passing through a fixed straight 
 line. 
 
 Let V=0 be the equation to the surface, then if its 
 tangent plane at wysz contain the straight line 
 
 A=mZ+a, Y=nZ +0, 
 we must have 
 dV dV adaV 
 
 — 0 
 Ce Tae 
 dV dV dV 
 — vw) —+(b-y)—-2—=0 (1). 
 (@- 0) T+ 0-5-8 5-9 O 
 
 But if V=w+v=0, so that w is the sum of all the terms of 
 m dimensions in V, we have by a property of homogeneous 
 functions 
 oes, Whe, du pros By 
 pad pote z—_= = = 
 da "dy dz 
 therefore equation (1) is reduced to 
 Puls Cael dv aay, 
 ———w | Sant av —- ——._- ———_—_- _ a se ad = 
 de dev dy - dy dz d 
 which is of the (z — 1)" degree; and the other two equations 
 from which together with (1) the co-ordinates of the points of 
 contact av, y, x are to be determined by elimination, are of the 
 n™ and (n —1)™ degrees; therefore the degree of the final 
 equation will generally be m(—1)*, which expresses the 
 number of the points of contact, or of tangent planes that can 
 
 be drawn through the fixed line. 
 
 159. In the preceding sections we have considered 
 several instances of surfaces generated by a line, straight or 
 curved, which so moves and changes its position or form, 
 as constantly to pass through one or more fixed curves or 
 directrices. We shall now extend the same considerations 
 
140 
 
 to the general case of a generating curve represented by the 
 equations 
 
 SCAR PA EN aah Pi Oy Opec eyed Ra 6 
 containing two variable parameters, a, $8, and subject to 
 pass through one fixed curve or directrix. The species 
 of the generating curve is determined, because the func- 
 tions f, f,, are supposed to be known, but its position and 
 dimensions will change corresponding to the different values 
 of a and 3; and it will generate a solid if the parameters 
 a and £ vary independently of one another. Thus the circle 
 represented by 
 
 +(y-By=ce-a’, v-a=0, 
 gives, by the elimination of a, the equation 
 S++ (y- Blac’, 
 
 which represents a sphere, radius c, and center in the axis 
 of y; and if £6 receive all values from 0 to + o, the equation 
 will belong to all the points of a solid cylinder, radius c, and 
 axis coinciding with the axis of y. 
 
 But if we suppose the generating line (1) in every position 
 to have a point in common with a fixed curve or directrix 
 represented by the equations 
 
 y= (#); s=vV(a), 
 
 then the equations to the generating line and directrix must 
 be simultaneously satisfied by the same system of values for 
 xv, y, * belonging to that common point; if therefore we 
 eliminate #, y, x between them, we shall have 3 = F(a), 
 the relation between the parameters, in order that the gene- 
 rating line may meet the directrix. Hence, the generating 
 line in any position will be represented by the system of 
 equations 
 
 S(® Ys % a, B)=0, fi(*, y, % a; 6B) =0, B= F(a). 
 
 If therefore we eliminate a and 8 between them, that is, 
 if we determine a and 3 from the two former in terms of 
 Xv, Y, %, so that B =u, a=v, and substitute in the latter, 
 
141 
 
 we shall have w= F'(v) for the equation to the surface; and 
 we observe that w and v do not change for surfaces of the 
 same family, that is, for those which admit the same gene- 
 .rating line, but that F', which depends upon ¢ and y, changes 
 for each Pcichion! surface of the family. 
 
 These considerations we shall now apply to several of the 
 more common cases, where the generating line is a straight 
 line, or a circle. 
 
 160. ‘To find the general equation to cylindrical sur- 
 faces. 
 
 A cylindrical surface is generated by a straight line 
 which moves parallel to itself, and always passes through 
 a given curve. 
 
 Let the equations to the generating line in any position 
 be v=mz+a, y=nzx +f, a and B being variable quantities 
 depending upon that position, and m and m constant quantities 
 since the line is always parallel to itself. Also let y= P (a), 
 #% =) (w), be the equations to the directrix or curve through 
 which the generating line always passes, and a’, y’, x’ the 
 co-ordinates of the point in which they meet; then a’, y’, 2 
 must satisfy both the equations to the directrix, and gene- 
 rating line ; 
 
 .a@ems ta, y=ne +B, y=O(v), x=’); 
 now by means of these four equations, we may eliminate 
 a’, y’, s, and there will remain 8 = F(a); but B=y- nz, 
 a=LX—-Ms; 
 
 . y—-nv=F(#- ms), 
 a relation among the co-ordinates of any point in the gene- 
 rating line, and therefore the equation to the surface which it 
 describes. In this case the quantities w and v are y — nz, and 
 « — mz, which remain the same for all cylinders, whilst the 
 function F' will alter with the different directrices employed. 
 
 Cor. If for the purpose of eliminating the arbitrary 
 function, we differentiate the equation y—nz = F'\(# — mz) 
 successively with respect to # and y, we find 
 
 —np=F'(4@—mz)(1-—mp), 1—-nq =F" (a — mz) (— mq) + 
 
142 
 
 therefore, dividing one result by the other, 
 
 np 1— mp dz dz 
 = > or m—+n— =I, 
 1—nq mq dx dy 
 
 which is the differential equation to cylindrical surfaces; we 
 may however obtain it more easily by the consideration of the 
 tangent plane, as follows. 
 
 161. To find the differential equation to cylindrical sur- 
 faces. 
 
 One of the distinguishing properties of cylindrical sur- 
 faces is, that the tangent plane, since it always contains a 
 generating line, (Cor. 2 Art. 108) is always parallel to a fixed 
 straight line. 
 
 Let 2 = ms’, y’=nz' be the equations to a line through 
 the origin, to which the generating line is always parallel; then 
 this line is parallel to the tangent plane whose equation is 
 
 s—s=p(e#—x#)+q(y'-y); 
 
 dz d 
 . (Art. 23) mp+nq=1, or mn Ere 1, as before. 
 
 Cor. This equation may be employed to discover whether 
 the surface represented by a proposed equation w = 0, is cylin- 
 
 drical or not. For, obtaining the values of p and q, as in Cor. 
 
 Art. 107, and substituting, we have 
 
 du du du 
 WP 9 0, 
 x z 
 
 which must be satisfied for all points of the surface, that is, 
 for all values of w, y, x, if the surface be cylindrical; hence 
 we must equate to zero the coefficients of the different powers 
 of the co-ordinates in it, and examine whether the resulting 
 conditions can be satisfied by real values of m and n. 
 
 162. Having given the direction of the generating line, 
 to find the equation to the cylindrical surface which envelopes 
 a given curve surface. 
 
143 
 
 We have seen that whatever be the nature of the directrix, 
 the equation mp+nq=1, must subsist between the differential 
 coefficients p, g, derived from the equation to a cylindrical 
 surface. But at the points where the cylinder touches the 
 surface, the values of p, g, are the same for both ; therefore at 
 those points the above relation subsists between the differential 
 coefficients derived from the equation to the surface; that is, 
 the co-ordinates of the points of contact are such that the 
 equation mp + nq =1 is satisfied. Hence if we differentiate the 
 given equation to the surface «=0, and write the values 
 thence obtained of p, g, in the above equation, the result 
 together with the equation to the surface, will be the equations 
 to the directrix ; and we know the direction of the generating 
 line, and therefore can find the equation to the required 
 surface by Art. 160, 
 
 163. To find the general equation to conical surfaces. 
 
 A. conical surface is generated by a straight line which 
 passes through a given point, and always meets a given curve. 
 
 Let the co-ordinates of the given point or vertex be 
 as Cs 
 
 @—-a=a(s—-c), y-b=B(s-0), 
 
 are the equations to the generating line in any position, a and 
 f being variable quantities depending upon that position. 
 Also let y = ¢ (#), % = W (@#), be the equations to the directrix 
 or curve through which the generating line always passes, 
 and a’, y’, x, the co-ordinates of the point in which they 
 meet; then a’, y’, x must satisfy both the equations to the 
 generating line and directrix, 
 
 wv —-a=a(%—-c),y -b=B(e'-0¢), y= (#), X= (e’). 
 
 Now by means of these four equations we may eliminate 
 we, y’, x, and there will remain 8 = F(a); 
 
 =U ioe 
 but 6 =~ sat ease ip 
 
 ? 
 yw — 6 w= 
 
14.4 
 
 J Y= (2) 
 
 Reb EG 
 
 a relation among the co-ordinates of any point in the generating 
 line, and therefore the equation to the surface which it de- 
 scribes. When the vertex is situated in the origin, the 
 equation is reduced to Pay (<) , which expresses that the 
 @ Pe 
 equation is homogeneous in w, y,%. It is easily seen that if the 
 directrix be the curve of intersection of two surfaces of the m™ 
 and 2" orders, the equation to the cone will be of the mn" order. 
 
 Cor. If from the above equation we eliminate the 
 arbitrary function by differentiation, as in Cor. Art. 160, 
 we find 
 
 tee d 
 zen 7 (@—a)t 7 yd), 
 
 which is the differential equation to conical surfaces; but 
 which may be obtained more easily by the consideration of 
 the tangent plane, as in the following Article. 
 
 164. To find the differential equation to conical surfaces, 
 The distinguishing property of conical surfaces is, that the 
 tangent plane, since it always contains one of the generating 
 lines (Cor. 2 Art. 108), always passes through the vertex. 
 
 Let a, b, c denote the co-ordinates of the vertex, then 
 they must satisfy the equation to the tangent plane at any 
 point wyz; that is, x —s=p(a —«x)+q(y'—y) must be 
 satisfied by # =a, y =b, x'=c; therefore, whatever be the 
 nature of the directrix, the differential coefficients p, g, derived 
 from the equation to the surface, must be such as to satisfy 
 the equation 
 
 x-c=p(w—a)+q(y—4), 
 
 which is the differential equation to conical surfaces. 
 
 Cor. As in cylindrical surfaces (Art. 161), this equation, 
 when put under the form 
 
 du du du 
 eae \ eee ny EMRE Cy ee 
 
145 
 
 may be employed to discover whether a proposed equation 
 u = 0 represents a conical surface or not; a, 6, ¢ being the 
 unknown quantities to which must be applied what was there 
 said relative to m and n. 
 
 165. Having given the position of the vertex, to find 
 the equation to the conical surface which envelopes a given 
 curve surface. 
 
 Whatever be the nature of the directrix, we have seen 
 that the differential coefficients derived from the equation to a 
 conical surface must satisfy the equation 
 
 s—-ce=p(w-a)+qty—); (1) 
 
 but at the points where the cone touches the surface, the 
 values of p,q, are the same for both; therefore at those 
 points the above relation subsists between the differential 
 coefficients derived from the equation to the surface, that 
 is, the co-ordinates of the points of contact are such, that 
 the above equation is satisfied. Hence if we obtain values 
 of p,q, from the given equation to the surface, and sub- 
 stitute them in the above equation, the result, together with 
 the equation to the surface, will be the equations to the 
 curve of contact of the two surfaces, or to the directrix ; 
 and we know the co-ordinates of the vertex, and can therefore 
 find the equation to the required conical surface by Art. 163 ; 
 and it will be of the m(m —1)™ order, if the given surface be 
 of the v" order, because (1) will become of the (7 — 1)" order, 
 in the same way as equation (1), p. 139. 
 
 Cor. 1. If the vertex of the cone be considered as a 
 luminous point, the curve of contact whose equations we 
 have just found, is that which on the surface separates the 
 illumined and obscure parts; if it be considered as the place 
 of the eye, the curve of contact is the line of the apparent 
 contour of the surface. 
 
 Cor. 2. Let w= 0 (1) be the equation to a surface of the 
 second order; then the curve of contact of the enveloping 
 cone is easily shewn to be a plane curve determined by an 
 equation of the first order, v =0 suppose. Hence w +2 .v* =0(2) 
 is another surface of the second order whose tangent cone 
 
 10 
 
146 
 
 from vertex (a, b, c) is identical with that to surface wu =0 
 from the same vertex, whatever be the value of the constant ) 3; 
 for from Art. 158 the surfaces (1) and (2) touch one another 
 along the curve in which they are intersected by v = 0, and 
 therefore each is touched by the same enveloping cone along 
 that curve. Now no surface of the second order can pass 
 through the vertex of its enveloping cone; for if it did, that 
 cone would become the tangent plane; therefore if > be 
 determined so as to make (2) pass through the vertex of the 
 cone, the equation w + Av” = 0, will represent the enveloping 
 cone of the surface wu = 0. 
 
 Similarly, if v =0 be the plane of contact of the envelop- 
 ing cylinder having its axis in a given direction; and 2d be 
 determined so as to render infinite that radius vector of the 
 surface % + \v* = 0 which is parallel to the axis of the cylinder, 
 we shall obtain the equation to the cylinder enveloping a 
 surface of the second order. 
 
 166. To find the general equation to conoidal surfaces. 
 
 A conoidal surface is generated by a straight line which 
 moves parallel to a given plane, and always meets a given 
 fixed straight line, and ‘a given curve. 
 
 Let the given plane, called the directing plane, be taken 
 for that of wy, and the point in which the straight directrix 
 OA (fig. 54) meets it, for the origin; then the equations to 
 OA will be r=mz, y=nz. Also let the equations to the 
 curvilinear directrix BC be y=@(@), s =W(a); and let 
 s=B, y=ax+- be the equations to the generating line 
 DC in any position, since it is always parallel to the plane 
 of vy; then since the generating line always meets OA, we 
 must have 
 
 nbB=amB+y; 
 
 therefore, eliminating y, the equations to the generating line 
 become 
 
 v= B, y-nB=a(w—mB); 
 
 we must next express that it meets the curve BC, whose 
 
147 
 
 equations are y=@(w#), x= (wv); therefore, eliminating 
 x,y, %, we have (3 = F(a); hence the generating line in any 
 position will be represented by the system of equations 
 
 e=B, y-nB=a(e-mp), B= F(a); 
 therefore, eliminating a and , the equation to the surface 
 generated is 
 
 g- (Zo). 
 
 C— MZ 
 
 Cor. 1. This surface is twisted, that is, no two consecu- 
 tive generating lines are in the same plane; for the line DC 
 in passing to the consecutive position D’C’ may be supposed 
 to glide along the tangent C7’; therefore, in order that DC 
 and D’C” may be in the same plane, OA and CT' must be in 
 the same plane, which cannot happen for all the tangents at 
 least, unless the curve BC is entirely in the same plane with 
 AO, in which case the surface generated will be a plane, 
 
 If the straight directrix be perpendicular to the directing 
 plane, m=0, m =0, and the equation to the surface, which 
 
 is then called a right conoid, becomes z = r(2); at which 
 
 we may readily arrive by a direct investigation, as the 
 equations to the generating line will be s =, y=aa. Also 
 it is manifest that the equation to the oblique conoid will 
 
 a ; : 
 assume the form z= r(2) if we refer it to a system of 
 
 & 
 oblique co-ordinate axes, of which OA is one, and any two 
 lines drawn through O in the directing plane are the others. 
 
 Cor, 2. If from the equation z = P(t) we elimi- 
 L—-Ms 
 
 nate the arbitrary function by differentiation, as in the 
 former cases, we find 
 
 p (w@—ms) + q(y— nz) =0, 
 
 the differential equation to conoidal surfaces; at which we 
 
 may arrive more easily as follows. 
 LO 
 
148 
 
 167. To find the differential equation to conoidal 
 surfaces. 
 
 Their characteristic property is that the tangent plane at 
 any point contains the generating line which passes through 
 that point, (for it contains the tangent lines to all curves 
 traced on the surface through that point, and a straight 
 line is its own tangent,) that is, a line parallel to the plane 
 of wy, and intersecting another line whose equations are 
 a =m, y’=nz'. Now the equation to the tangent plane at 
 a point vyz is 
 
 x —s=p(a-x)+q(y'-y), 
 
 and if we intersect it by a plane parallel to ry at a dis- 
 tance x from it, their line of intersection will have for its 
 equations 
 
 s'=s, p(w—x)+q(y-y) =0, 
 and since this is coincident with the generating line, it will 
 intersect the rectilinear directrix of which the equations are 
 wv =m, y=ne'; 
 
  p(w-—msz)+q(y—nz)=0, 
 
 which is the relation that must be satisfied by the co-ordinates 
 of every point in the surface. 
 
 For a right conoid, or for an oblique conoid referred 
 to the rectilinear directrix as one of the axes, the differential 
 equation is 
 
 p“e+qy=d0. 
 
 168. Having given the rectilinear directrix, to find the 
 equation to the conoidal surface which envelopes a given 
 curve surface. 
 
 As in the former cases, if %=0 be the equation to the 
 given surface, the curve of contact which forms the curvi- 
 linear directrix, will be given by the equations 
 
 du du 
 0, Cl atisted eg 88) ED arte 
 
149 
 
 which must be employed instead of the equations y = ¢ (2), 
 x =wW(w#); the remainder of the process will be the same 
 
 as that explained in Art 166. 
 
 169. To find the general equation to surfaces of re- 
 volution. 
 
 A surface of revolution is generated by a curve revolving 
 about a fixed axis. If the curve be situated entirely in a 
 plane passing through the axis, then every section of the 
 surface through the axis, which is called a meridian, will 
 reproduce the generating curve; otherwise, the section will 
 be different from the generating curve. Also every section 
 perpendicular to the axis, which is called a parallel of the 
 surface, will be a circle whose center is in the axis and 
 plane perpendicular to it, and whose perimeter passes through 
 the generating curve; hence a surface of revolution may be 
 supposed to be generated by a circle whose center moves 
 along a fixed straight line to which its plane is perpendicular, 
 and whose perimeter always passes through a given curve. 
 The advantages of the second mode of generation are, that 
 the moveable circle is a generating curve of an invariable 
 nature and common to all surfaces of the class, whilst the 
 curve through which it always passes is a variable directrix 
 which changes for each particular surface, agreeably to the 
 mode of generation occurring in every instance in this section. 
 
 Let v-—-a=m(z-c), y-b=n(s—- Cc), be the equations 
 to the axis or fixed straight line along which the center of 
 the generating circle moves, and which we suppose to be 
 drawn through a given point (a, 6, c) in a known direction. 
 Then every one of the parallels of the surface may be re- 
 garded as the intersection of a plane perpendicular to this 
 axis, with a sphere whose center is any fixed point in the 
 axis, the point (a, b, c), for instance, and whose radius varies 
 so as to give the section of the proper magnitude; conse- 
 quently the equations to the generating circle will be 
 
 Neate () 
 (vw — a)? + (y — b)? + (ew — cp whis,0 ate ; 
 
 ll 
 7 
 
150 
 
 Let y = ¢ (x), = = (2), be the equations to the directrix 
 or curve which the perimeter of the generating circle always 
 meets; then these four equations will be satisfied by the same 
 system of values of #, y, x, which are the co-ordinates of the 
 point of intersection; hence, eliminating a, y, x from them, 
 we find B = F(a) the condition to be joined to the equations 
 (1) in order that the circle which they represent may really 
 be a parallel of the surface; hence, eliminating a and 3 from 
 the equations to the parallel, we find 
 
 s+me+ny =F }(e—a) + (yb)? + (z- 6), 
 
 a relation among the co-ordinates of the generating circle in 
 any position, and therefore the equation to the surface which 
 it describes. 
 
 Cor. 1. If the axis of revolution coincide with the axis 
 of zg, and we take the center of the sphere for origin, we 
 have 
 
 m=n=0, a=b=c=0, and z= F(a#+y¥ + 2°), 
 
 or x= f(a" +y’), 
 as we have already seen, Art. 52. In this case it is more 
 convenient to regard each parallel as the intersection of a 
 right cylinder with a plane perpendicular to it, that is, 
 instead of equations (1), to use 
 
 = B, a +y’ =a, 
 
 which joined to the usual relation 6 = F(a), give imme- 
 diately 
 
 s= F(a’ + y’). 
 
 Cor. 2. If we differentiate the above equation with 
 respect to w and y, and divide one result by the other, 
 so as to eliminate the arbitrary function, we find 
 
 m+p «w-a+(x—c)p 
 m+q y-b+(s—e)q’ 
 or pjy—b—n(z—c) —q}w-a—m(z-c)} =n(w—a)—m(y-b), 
 
 the differential equation to surfaces of revolution, at which 
 we may also arrive as follows. 
 
151 
 
 170. To find the differential equation to surfaces of 
 revolution. 
 
 Their distinguishing property is, that the normal always 
 meets the axis of revolution. This will be readily seen if 
 we consider that the tangent plane at any point necessarily 
 contains the line touching the parallel at that point, and is 
 therefore perpendicular to the meridian plane passing through 
 that point; consequently the normal will lie in the meridian 
 plane, and therefore meet the axis. 
 
 The equations to the normal at a point wyxz are 
 eo —-ve+p(2'-x)=0, y-y+q(e' -2) =0; 
 let, as above, the equations to the axis be 
 wo —-a—m(x —c)=0, y —b-—n(s'-c)=0, 
 then if these two lines intersect, their equations must be 
 satisfied by the same values of a’, y’, 2’, which will be 
 the co-ordinates of their point of intersection. Hence if we 
 eliminate a’, y’, 2’, the result will be the differential equation 
 to the surface. Subtracting, we find 
 m (x'—c)+p(2’-2) -@ +a=0, n(x’ —c) +q(%—2)-yt+b=0, 
 or, (m +p) (x —2) +m(x—c) -v%+a=0, 
 (n +q)(s —s) +n (e@-c)-yt+b=0; 
 
 -, eliminating x’ — x, we find as before 
 
 (m+ p) {n (# —0) —y +b} = (m4 q) {m(w—c)-w+a}...(1); 
 which expresses that a surface is one of revolution about an 
 axis passing through a point a, 6, c, and having a direction 
 determined by the quantities m and 7. 
 
 If the axis of revolution coincide with the axis of z, the 
 equation becomes 
 
 py—-gqr=d0. 
 
 Cor. The above equation will enable us to ascertain 
 whether a proposed equation w=0, represents a surface 
 of revolution; fer, deducing the values of p and q, and 
 substituting them in (1), the result ought to be true for 
 
152 
 
 all values of w, y, x. Hence we must equate to zero the 
 coefficients of the different powers of the co-ordinates, and 
 we shall obtain among the unknown quantities a, b, c, m, n 
 a certain number of equations, which must be consistent 
 with one another, in order that the surface may be one of 
 revolution. The process applied to the general equation of 
 the second order will furnish the conditions already obtained 
 (Art. 149) by a different method. 
 
 171. Having given the position of the axis, to find the 
 equation to the surface of revolution which shall envelope a 
 given curve surface. 
 
 As in the similar cases which have preceded, we must 
 begin by finding the equations to the curve of contact 
 which will be the directrix of the moveable circle. Let 
 wz=0 be the equation to the proposed surface, then at all 
 points along the curve of contact the tangent planes of the 
 two surfaces are coincident; and therefore the values of p 
 and q derived from w = 0, and substituted in (1), must satisfy 
 it for all points in the curve of contact; the result of this 
 substitution together with w= 0, will therefore be the equa- 
 tions to the directrix, and then the equation to the required 
 surface of revolution may be found as before. 
 
 It is manifest that we thus determine a surface which, 
 if the proposed surface revolved about the given axis to 
 which it was fixed, would touch and envelope the proposed 
 surface in every position. 
 
 172. The above are the principal propositions relative 
 to the commoner families of surfaces which admit only one 
 directrix, or only one of a variable form; for though Conoidal 
 surfaces have two directrices, yet one of them, namely the 
 rectilinear one, is constant for all surfaces of the family. 
 Consequently, for Conoidal, as well as for all the other 
 families of surfaces considered in this section, the finite 
 equation involves only one arbitrary function, and the dif- 
 ferential equation cleared of the arbitrary function, is only 
 of the first order. 
 
SECTION VIII. 
 
 ON SURFACES HAVING MORE THAN ONE ARBITRARY CURVI- 
 LINEAR DIRECTRIX, AND ON DEVELOPABLE SURFACES, AND 
 ENVELOPES, 
 
 173. In the preceding Section we considered several 
 surfaces where the equations to the generating curve con- 
 tained two variable parameters; and the necessary dependence 
 of these parameters upon one another, in order that a surface 
 and not a solid might be generated, was established by making 
 the generating curve in every position have a point in common 
 with a given fixed directrix. The resulting equation to the 
 surface consequently involved but one arbitrary function; and 
 the differential equation, characterizing the nature of the 
 surface independent of that directrix, was of the first order. 
 
 We shall now take a more general view of the subject of 
 the generation of surfaces, and suppose the equations to 
 the generating curve 
 
 St (@, Yo By As Bs Y &e.) = 0, fi (a, Y, Bs a, B; ie &c.) = 0, 
 
 to contain any number 7 of variable parameters a, 9, y, &c. 
 This curve will not, by passing through all the various 
 positions and forms corresponding to the various values of 
 the parameters, generate a determinate surface, unless we 
 subject it to such conditions as will leave only one of the 
 parameters, a for instance, arbitrary ; and this may be done 
 by making it, in every position, pass through m —1 curves 
 represented by the equations 
 
 y=P(%) F=W(*%)s Y= hilt) x= VWn(a); &e. 
 
 For in order that it may have a point in common with the 
 first directrix, its two equations and the equations y = ¢ («) 
 z= (a) must be satisfied by the same system of values 
 of w, y, x, which are the co-ordinates of that point; there- 
 fore, eliminating w, y, * between these four equations, we 
 
154 
 
 find F(a, 3, y, &c.)=0, which expresses the necessary 
 dependence among the parameters in order that the gene- 
 rating curve may pass through the first directrix. In the 
 same manner, each directrix will furnish a fresh relation 
 between a, 3, yy, &c.; so that to represent completely the 
 generating curve passing through the 2 — 1 directrices, we 
 must join to its two equations, the 2 —1 equations of con- 
 dition F' (a, B, y, &c.) = 0, F, (a, B, y, &c.) = 0, &c. And 
 if from these latter we suppose $B, y, &c. to be all de- 
 termined in terms of a, and substitute their values in the 
 equations to the generating curve, those equations will then 
 involve only one parameter; and if we finally eliminate that 
 parameter between them, we shall arrive at an equation con- 
 taining no arbitrary quantity and representing a determinate 
 surface generated by the moveable curve. 
 
 Hence the final equation will contain as many arbitrary 
 functions as there are directrices, and therefore the differential 
 equation characterizing the surface independent of the di- 
 rectrices and consequently not involving the functions, will be 
 of an elevated order generally exceeding the number  — 1 of 
 functions, and never less than 22—3. ‘These principles we 
 shall now illustrate, confining ourselves, however, to the cases 
 where the generating curve is a straight line. 
 
 174. Surfaces generated by a straight line are divisible 
 into two classes, each of which has distinct properties; viz. 
 into twisted surfaces, where two consecutive positions of the 
 generating line are never in the same plane; and into de- 
 velopable surfaces, where two consecutive positions of the 
 generating line are always in the same plane. 
 
 One remarkable difference in the properties of these two 
 classes has reference to their contact with the tangent plane ; 
 in both, the tangent plane at any point (since it contains 
 the tangent lines to all curves traced on the surface through 
 that point, and a straight line is its own tangent) must 
 contain the generating line passing through that point; but 
 in twisted surfaces, the tangent planes along a generating 
 line are all distinct and each touches the surface only in 
 
155 
 
 a point; whereas in developable surfaces, it is one and the 
 same plane which touches the surface at all points along 
 a generating line. 
 
 For let DC, DC’ (fig. 54) be consecutive positions of 
 the generating line of a twisted surface; BC, ED sections 
 of the surface through any points D and C, and TC, AD 
 tangent lines to those sections. Then the tangent planes 
 at all points along DC contain DC; and at D and C the 
 tangent planes are ADC, T'CD, which cannot be coincident 
 unless AD, T'C are in the same plane; but in that case D’C’ 
 (since between the positions DC, D’'C’, the generating line 
 may be supposed to glide along 4D, CT’) would be in the 
 same plane with DC, which is impossible since the surface 
 is twisted. But if the surface be developable, then since 
 D'C’ is in the same plane with DC, the tangents 4D, TC 
 are in the same plane, and therefore the tangent planes at 
 D and C, as well as at every other point along DC, are 
 coincident. This property has already been proved for 
 cylinders and cones, which are evidently particular cases of 
 developable surfaces. 
 
 175. We shall first consider twisted surfaces; and 
 among the various ways in which the motion of a straight 
 line may be governed so as to generate a twisted surface, 
 we shall select the two following: viz. when the moveable 
 line constantly passes through two fixed curves and remains 
 parallel to a fixed plane, and when it constantly passes 
 through three fixed curves. 
 
 176. To find the equation to the surface generated by 
 a straight line which constantly passes through two given 
 curves, and remains parallel to a fixed plane. 
 
 The motion of the generating line is here, manifestly, 
 completely determined, as we have only to cut the curves 
 by planes parallel to the fixed plane, and to join the points 
 of section by straight lines, to obtain as many positions of 
 the generating line as we please; and as the tangents to the 
 
156 
 
 fixed curves at those points of section will not usually be in 
 the same plane, the surface will generally be twisted. 
 
 Taking the directing plane for that of ay, let the equa- 
 tions to the generating line in any position be 
 S=a, y=PBat+y, 
 and let the equations to the two directrices be 
 y= oie) #=\,(0)3 y= dua) = yA (0); 
 then successively combining the equations to the moveable 
 
 line with those to each of the directrices, so as to eliminate 
 VY, Y, %, we find 
 
 F, (a, B, y) = 0, 1 Gd (rey decry Lal Oe 
 
 which we may suppose reduced to the forms 
 
 B = p(a), y = W (a). 
 
 Hence, eliminating a, 3, y between these and the equations 
 to the generating line, we have for the equation to the surface 
 
 y = vp (zx) +(e). 
 
 Cor. ‘To find the differential equation, we have, differen- 
 tiating successively with respect to w and y, and dividing one 
 result by the other, 
 
 0= (2) + 2p) p+ ¥'()p, 
 1=0$' (2) 9+ V4 
 
 2 oh ea 
 
 (=~ 9G) 
 
 therefore, differentiating again, and dividing one result by the 
 other, employing the usual notation, 
 
 qr—ps yee qs — pt j 
 g =- (2) p, ny =— (2) q; 
 
 “ Fr—2pqst+ pt =0, 
 
 the equation common to all surfaces of this class. 
 
157 
 
 177. To find the equation to the surface generated by 
 a straight line subject to constantly pass through three given 
 fixed curves. 3 
 
 It is easily seen that the motion of the generating line 
 will be completely determined. For conceive any point in 
 one of the fixed curves to be the common vertex of two 
 conical surfaces, having the other two fixed curves for their 
 directrices; then these surfaces will intersect one another in 
 a finite number of straight lines, each of which passes through 
 the three fixed curves and is therefore a position of the gene- 
 rating line; and as we take fresh points for the vertex, the 
 successive generating lines belonging to the same sheet will 
 pass through points contiguous to one another on the three 
 directrices. Also the surface will generally be twisted, because 
 the tangents to the three directrices at the points where a 
 generating line meets them, will not, except in very particular 
 cases, be in the same plane. 
 
 Let the equations to the generating line be 
 w=axsty, y=Bz +o, 
 
 containing four variable parameters, a, (6, y, Oo; and, as 
 explained in Art. 173, let the three relations among the 
 parameters necessary for the line passing through the three 
 given directrices, be obtained and reduced to the form 
 
 B=9(2), y=) d=7(a)s 
 
 therefore, substituting for 8, y, 0 in the equations to the 
 generating line, we have 
 
 w=azt+W(a), y=xh(a) +7 (a); 
 
 and it remains to eliminate a between these equations, in 
 order to get the equation to the surface ; but this cannot 
 be done without particularizing the functions, or the direc- 
 trices on which they depend. ‘Therefore we must retain the 
 above system of equations to represent this family of surfaces ; 
 regarding a as an indeterminate quantity to be eliminated, 
 after the forms of the functions in each individual case have 
 been determined. 
 
158 
 
 It may be observed that the various, other modes of gene- 
 rating this sort of surface, as for instance when we replace one 
 or more of the directrices by a surface which the moveable 
 line is always to touch, are reducible to this method; for in 
 every case we may take for the three directrices, any three 
 sections of the generated surface made at pleasure. 
 
 Cor. To obtain the differential equation independent 
 of the directrices, we must, as before, obtain by successive 
 differentiations of the above system a sufficient number of 
 equations to eliminate a, and the functions @, vy, mw and their 
 derived functions; the result is a complicated equation of 
 the third order. 
 
 When in the two preceding cases, the directrices all 
 become straight lines, the surfaces generated become re- 
 spectively the hyperbolic paraboloid, and the hyperboloid 
 of one sheet, as is seen in the Appendix; they are the only 
 twisted surfaces whose equations do not rise above the second 
 degree. 
 
 Developable Surfaces. 
 
 178. We next come to the consideration of the second 
 class of surfaces which admit of being generated by a straight 
 line, the characteristic law of the motion being that two 
 consecutive positions are always in the same plane. Before 
 proceeding to point out the various modes in which this 
 condition may be satisfied, we shall shew that surfaces ge- 
 nerated in this manner are developable; that is, supposing 
 them flexible but inextensible, they may without rumpling 
 or tearing be made to coincide with a plane in all their 
 points. 
 
 Let fig. 57 represent a surface of this sort, and let AN, 
 A'N’, A”N”, &c. be positions of the generating line inde- 
 finitely near to one another; then from the definition of the 
 surface, 4N will be intersected by A’N’ in some point m, 
 A'N’ by A’ N” in m’, and the latter by the next generating 
 line in m", and so on; so that these successive points of 
 
159 
 
 intersection will form a polygon mm’m’”..., or rather a con- 
 tinuous curve to which all the generating lines are tangents. 
 Also AN and A’N’, and similarly every pair of consecutive 
 generating lines, will include a sectorial area Am A’ of inde- 
 finite length, but infinitely small angle, which may be re- 
 garded as a plane element of the surface. If now the first 
 of these elements be turned about its line of intersection with 
 the second, till they are in the same plane; and then the 
 system formed by these two be turned about the line of 
 intersection of the second and third till they are in the same 
 plane; and if this operation be continued through all the 
 plane elements, they will all thus be brought into one plane, 
 and the given surface will be developed without rumpling or 
 tearing. 
 
 179. We have already seen (Art. 174) that the plane 
 which touches a developable surface in any point M’ is the 
 tangent plane at every point in the generating line passing 
 through JZ’; and that to construct the tangent plane to any 
 point 1’, we have only to draw a tangent line M’T' to any 
 curve on the surface passing through that point, then TM'N’' is 
 the tangent plane required. 
 
 180. As the generating lines are all tangents to the 
 curve mm'm’,.. formed by their perpetual intersection, the 
 surface may be supposed to be generated by a moveable 
 straight line which is always a tangent to a fixed curve; 
 the curve must of course be of double curvature, otherwise 
 the surface generated would be a plane. Hence it is suf- 
 ficient to assign one fixed directrix (to which the generating 
 line must be always a tangent) to completely determine a 
 developable surface. 
 
 If the equations to the fixed curve be 
 
 v= (2%) y=), 
 
 the equation to the surface will result from eliminating a from 
 the equations to the line touching the curve at a point for 
 which s =.a, which (Art. 123) are 
 
 2 $@)=$@MC-@, ¥-V¥O=V@(e- 45 
 
160 
 
 the elimination cannot be effected without assigning the 
 forms of the functions; but every developable surface may 
 be represented by the above system of equations, regarding 
 a in the first as a function of y and x to be determined from 
 the second. 
 
 181. The curve mm'm”... is called the edge of re- 
 gression of the surface which its tangent generates; the 
 reason of which is that the two portions of any tangent, 
 produced both ways from the point of contact, generate two 
 distinct sheets of the surface which are united in the curve; 
 so that if we wished to pass from a point in one sheet to 
 a point in the other without quitting the surface, our path 
 would have a point of regression at the curve, provided the 
 two points were not in the same generating line. In conical 
 surfaces the edge of regression is reduced to a point, in 
 cylindrical surfaces it 1s removed to an infinite distance. 
 
 182. We may also generate a developable surface, by 
 assigning two fixed directrices through which the moveable 
 line is always to pass; in that case the points in which 
 the moveable line meets them must be constantly so chosen 
 that the tangent lines to the directrices at those points are 
 in the same plane; then, as the moveable line in passing 
 into its consecutive position may be supposed to glide along 
 those tangents, every two consecutive positions will be in 
 the same plane. Consequently, after having obtained two 
 relations among the parameters which enter into the equa- 
 tions to the moveable line by subjecting it to meet the 
 two directrices, we shall obtain a third by expressing that 
 the tangents to the directrices at the points where the 
 moveable line meets them, are in the same plane; hence 
 only one arbitrary parameter will remain in the equations, 
 by eliminating which we shall obtain the equation to the 
 surface. 
 
 183." There is yet another way of expressing the gene- 
 ration of developable surfaces; for since every two consecutive 
 generating lines include between them a plane sectorial area 
 
161 
 
 of infinite length, we may regard these elements as forming 
 parts of the successive positions of a plane subject to move 
 according to a given law. This method will usually be found 
 the most convenient in practical applications; and it leads 
 easily to the general differential equation to developable 
 surfaces, and to other results, which we shall now obtain by 
 means of it. 
 
 184, To find the general equation to developable sur- 
 faces, considered as generated by the consecutive intersections 
 of the positions assumed by a plane subject to move after 
 a given law. 
 
 The law according to which the plane moves will vary 
 with each particular surface; but in order that the motion 
 may be completely determined, and that a single surface 
 and not an infinite number of surfaces may be generated, 
 it must leave only one arbitrary parameter in the equation 
 to the moveable plane. The successive positions assumed 
 by the plane by virtue of the infinitely small variations of 
 the parameter, will cut one another consecutively in straight 
 lines, which taken two and two are in the same plane; 
 these lines will therefore form a developable surface touched 
 by all the planes. Hence the law of succession of the planes 
 which generate a developable surface requires that two of 
 the three arbitrary constants which enter into the equation 
 to a plane should be functions of the third, or, which is the 
 same thing, that they should all be functions of the same 
 parameter a; let therefore 
 
 s=ap(a) + yf(a) + (a) 
 
 be the equation to the generating plane in one of its positions 
 depending upon the parameter a; then the equation to the 
 plane, which differs insensibly from it in position, will be 
 
 ce an) Sat &et + yf f(a) +f (a). da + &c.} 
 | +W(a) + W (a). Sa + &e.; 
 
 and the co-ordinates of the points in which they intersect 
 LL 
 
162 
 
 must satify these two equations; that is, they must satisfy 
 z= 0(a) +yf (a) + (a), 
 0O= w$ (a) + &e.t + y sf’ (a) + &e.$ + W' (a) + &e. 5 
 
 or if the planes be consecutive, making da = 0, the equations 
 to their line of intersection are 
 
 s=adp(a)+yf(a) + (a)...... Gly. 
 O0=ad'(a)+yf (a)+ (a)... (2); 
 
 and the general equation to developable surfaces will result 
 from eliminating a between these equations; but as the 
 elimination cannot be effected without fixing the forms of the 
 functions ¢, f, .y, that is, without particularizing the surface, 
 we must retain the above system to represent this family 
 of surfaces, regarding a in the former as a function of w and y 
 to be determined by the latter. 
 
 Cor. Hence the differential equation to developable 
 surfaces can be obtained; for, differentiating (1) successively 
 with respect to w and y, regarding a as a function of # and y 
 to be determined from (2), we have 
 
 p=$(a) + feg'(a) uf) +O} S. 
 
 vis prin’ el ibsidetta 
 q=f(a) + {2p'(@) + of (a) +V(a@)} ar 
 which by virtue of equation (2) are reduced to 
 
 p= (a), g=f(a); 
 
 and since p and q are functions of the same quantity, they 
 are functions of one another, 
 
 “p= 7 (q); 
 
 which is the differential equation of the first order to de- 
 velopable surfaces, containing one arbitrary function; this 
 may be made to disappear by differentiating the equation 
 
163 
 
 p = m(q) successively with respect to # and y, and dividing 
 one result by the other; for we have 
 
 =a (q).s, 8=(q)-é; 
 .rt—s=0, 
 
 the differential equation of the second order common to all 
 developable surfaces. 
 
 185. To find the equations to the edge of regression 
 of a developable surface. 
 
 The edge of regression is the locus of the intersections 
 of the successive generating lines; and to find its equations we 
 must combine the equations to any generating line (1) and (2), 
 in which a is an arbitrary constant, with the equations to the 
 consecutive generating line resulting from (1) and (2), by 
 changing a into a + 0a; we shall thus obtain four equations 
 which by reduction are equivalent to the three 
 
 z= ad(a)+ yf (a) + W (a), 
 
 0=ag (a) + yf (a) + ¥ (a), 
 
 0= ag" (a) +f (a) + W' (a); 
 these, for any value of a, will furnish the three co-ordinates 
 x,y, % of the point where the generating line corresponding 
 to that value of a, is intersected by the consecutive generating 
 line; so that in eliminating a from the above system, we 
 shall have the two equations to the locus of all such points of 
 intersection; that is, to the edge of regression of the surface, 
 
 by the tangent line to which the surface may be considered 
 as generated. 
 
 186. ‘To find the equation to the developable surface 
 which touches at the same time two given curve surfaces. 
 
 Let the equations to the two surfaces be 
 F, (a, Yis %1) ee Ag ee, (#5, Y25 Bo) = 0, 
 
 dz dz 
 
 fie 
 dase tt dy 
 
 11—2 
 
 and let p,4i, P2q2, denote the values of for the 
 
164 
 
 first and second surface respectively. Then the moveable plane, 
 since it touches the first surface, will have for its equation 
 
 B—-R= pi (©-X)+UnY-Y%); 
 and in order that it may touch the second surface, that is, 
 in order that it may coincide with the tangent plane at 
 the unknown point «,y,%,, we must join the conditions 
 
 Pi = Por Nh = Joo F1 — PiG — NY = Fe — P22 — G2Yr2- 
 If therefore from these three equations, and the two equations 
 to the surfaces, we determine the five unknown quantities 
 
 Yis Zip Voy Yoo Vy in terms of &,, the equation to the moveable 
 plane will take the form 
 
 x =ad(a) + yf(a) + (a); 
 
 and by eliminating the arbitrary quantity #, between this 
 equation and the derived equation with respect to v,, we shall 
 obtain the equation to the required developable surface. 
 
 If we suppose an opaque body to be illuminated by a 
 luminous one, the surfaces which circumscribe the umbra and 
 penumbra occasioned by the interposition of the opaque body, 
 are two sheets of the developable surface which touches 
 both the bodies; and its lines of contact with the surfaces 
 of the bodies are, one, the curve on the opaque body which 
 separates the illumined and obscure parts; and the other, 
 the curve on the luminous body which separates the illumi- 
 nating part from that which can send no rays to the other 
 body. Consequently this problem is the general problem of 
 
 umbras and penum bras. 
 
 Similarly, we may find the developable surface generated 
 by the intersections of the tangent planes to a surface along an 
 assigned curve traced on the surface. 
 
 187. We may arrive at the differential equation to de- 
 velopable surfaces, and at the equation to the generating line 
 passing through any point, by the following method. 
 
 Since a tangent plane applied to the surface at any point, 
 touches it in a series of points the projection of which on any 
 
165 
 
 of the co-ordinate planes is a straight line, and since the 
 equation to the tangent plane at a point wysx is 
 
 = pa +qy +8 —pxr—qy, 
 if y’ =ma' +n be the equation to the projection of the ge- 
 nerating line which passes through the point wy, and in 
 the values of p, g, and s — paw —qy, we change y into ma + n, 
 the above equation must remain the same for all values of 
 xv whatever, and therefore the quantities p, q, 7 — px —qy, 
 which become functions of # only, must have their differential 
 coefficients with respect to w equal to nothing, 
 
 dp dy 
 _— = —~=0 
 dw 4 u ane: ; 
 d d 
 q =s+f y = 0, 
 dav dav 
 
 d 1733 4 BS ( + 
 FMS aiN a WP Gar HP Oe 
 
 te sg : 
 Hence, eliminating as rt —s* =0, the equation required ; 
 
 x 
 : . s ; 
 and y -—y=— 5 (# —uv)y=-— - (x — x) is the equation to the 
 
 projection of the generating line which passes through the 
 point xy%. 
 
 188. When a developable surface is made plane, the 
 absolute lengths of any determinate portions of the gene- 
 rating lines, as well as of any curve traced on the surface, 
 are not altered; and the angle which the tangent line to the 
 curve at any point forms with the generating line through 
 that point also remains unaltered ; but the curvature at any 
 point is changed. 
 
 For let AN, A’N’, &c. (fig. 57) be consecutive positions 
 of the generating line of a developable surface, and let PR, 
 P’R, &c. be the chords produced of a curve traced on this 
 surface, and meeting the generating lines in the points P, P’, 
 &c.; and for the surface let us substitute the series of plane 
 sectorial elements Am JA’, A’m’ A”, &c., and for the curve the 
 
166 
 
 polygon PP’P” &c. Then in turning the first plane element 
 about A’m to bring it into the same plane with the second, 
 and in turning the system formed by the two about A”m’ to 
 bring it into the same plane with the third, and so on, it is 
 evident that neither the lengths of PN, P’N’, &c., nor the 
 angles which they form with PP’, P’P’, &c., nor the lengths 
 of the latter, are altered, but that the angles RP’R’, RP’ R", 
 &c. will be changed; and as this continues true, however 
 small we take the plane elements, the properties announced 
 above for the surface and the curve traced upon it, are 
 
 established. 
 
 189. The shortest line on a developable surface has the 
 property that its osculating plane at every point is perpen- 
 dicular to the plane touching the surface at that point, or 
 contains the normal to the surface at that point. 
 
 If the polygon be such that, upon bringing all the plane 
 elements into the same plane, it becomes a straight line, then 
 two consecutive sides must always make the same angle with 
 the intermediate generating line; that is, for every point we 
 must have 
 
 L4mPR=em PR; 
 
 for when this condition is fulfilled, in the developed surface 
 every two consecutive sides will be the prolongment of one 
 another. Also since RP’, R’P’, make equal angles with 
 P'N', they may be regarded as two generating lines of a 
 conical surface whose axis is P’N’, and therefore the plane 
 RP'R’ will ultimately be the tangent plane to this surface 
 along P’R, and therefore perpendicular to RP’N’, which is a 
 meridian plane of the cone containing P’R; hence, passing 
 to the curve, the plane RP’R’ is its osculating plane at P’, 
 and RP'N’ is the tangent plane to the surface at the same 
 point, and these planes are perpendicular to one another; 
 also since, when the surface is developed, the curve becomes 
 a straight line, it is the shortest line which can connect two 
 points on the surface through which it passes, and it has the 
 property announced at every point. 
 
167 
 
 190. That the same property is true for the shortest 
 line traced on any surface whatever, appears thus. 
 
 If on any proposed surface we conceive the shortest line 
 between any two points to be traced, we may describe a 
 developable surface which shall touch the surface according 
 to that curve; consequently the curve of contact will be the 
 shortest line between the same points on the developable 
 surface, and therefore its osculating plane at every point must 
 contain the normal to the developable surface, i. e. the 
 normal to the proposed surface at that point. Hence if 
 y = d (x), x = yy (x) be the unknown equations to the shortest 
 
 ; d 
 line on a surface for which = 7, - =q; expressing that 
 
 the osculating plane at any point contains the normal to the 
 surface at that point, we find 
 
 dy (1 =) d’sf dy ); 
 dx? beers = Sag -4 
 
 d d 
 which together with e =p+t+ 7 will determine the curve. 
 v c 
 
 If we take s the length of the arc for independent variable, the 
 equations to the curve will assume the symmetrical forms 
 ax d’z d’y d’z 
 
 = : a 
 ds” Leryn ?- ds? os Tae 
 
 0. 
 
 191. From the property that the lengths of the ge- 
 nerating lines of a developable surface, and of any curve 
 traced upon it, as also the angles which the generating lines 
 form with the curve, are not altered by the development of 
 the surface, we can find the nature of the curve when the 
 surface is made plane; or, conversely, a curve being traced 
 upon a plane, we can find its nature when the plane is applied 
 to a given developable surface. This is shewn in the case of 
 cylindrical and conical surfaces, in the appendix. 
 
 Envelopes. 
 
 192. Developable surfaces, considered as generated by 
 the successive intersections of planes drawn after a given law, 
 
168 
 
 are a particular case of a general family of surfaces to which 
 we shall in the next place direct our attention. 
 
 193. To find the equation to a surface which envelopes 
 a series of surfaces described after a given law. 
 
 Let w=f(x, y, %, a)=0 be the equation to a surface 
 containing, besides other constants, a parameter a; then if 
 we give a particular value to a, the form and position of 
 the surface in space will be completely determined; and if 
 we give to it all possible consecutive values, we shall obtain 
 an infinite number of corresponding surfaces, usually inter- 
 secting one another two and two. The surface formed by 
 these successive intersections, and having with each one of 
 the surfaces the infinitely narrow zone in common which is 
 contained between the curves in which that individual surface 
 is intersected by the preceding and succeeding ones, (and 
 which consequently envelopes each one of the first series of 
 surfaces, supposing them all to exist together, and touches 
 it according to a curve of intersection) has been named by 
 Monge the Envelope. Also the curve in which any two 
 consecutive surfaces intersect, he has named the Characteristic 
 of the Envelope, because it indicates the mode of its genera- 
 tion; thus the characteristic of developable surfaces which 
 are generated by the intersections of planes, is a straight line ; 
 and that of surfaces generated by the intersections of spheres, 
 a circle, : 
 
 To find the equation to the envelope, we observe that if 
 after having given the parameter a determined value a in 
 u = 0, we give it a new value a + oa differing insensibly from 
 the former, we obtain the equation to a second surface differing 
 in form and position insensibly from the first, and intersecting 
 it in a series of potnts the co-ordinates of which satisfy the 
 equations 
 
 du . @u (da)® 
 
 = 0, U-+ ——.0a + ——. 
 da d a’ 24 
 
 du 
 or u=0, apes ee 
 a 
 
169 
 
 or if the surfaces be consecutive, w= 0, — =0;3 
 
 it follows therefore that the equations to the characteristic or 
 curve of intersection of two consecutive surfaces, corresponding 
 to the value a of the parameter, are 
 
 d 
 S(@, Y, % a) =9, Spake Y,.% a) =0;3 
 
 and since the envelope is formed by the assemblage of the 
 characteristics, its equation will result from eliminating a 
 between the same equations. As this elimination, however, 
 cannot be effected without assigning the form of f; we must 
 retain the above system of equations to represent the envelope, 
 regarding a in the former as a function of w, y and z to 
 be determined from the latter. 
 
 194, Again, if in the equations to the characteristic, 
 after having given the parameter a particular value a, which 
 determines the position of the characteristic in space, we give 
 it a new and insensibly increased value a+ da, the two 
 resulting equations will be those to a second characteristic, 
 differing insensibly in form and position from the first, and 
 intersecting it in general in a finite number of points, the 
 co-ordinates of which satisfy the two sets of equations 
 
 du 
 u = 0, ory 
 du du du 
 —. ae U) pa ——— , eee 
 Uti da + &c. = 0, FEEL da + &e : 
 
 which latter two equations, by virtue of the former, are 
 equivalent to 
 
 du & " ad’ u 
 rere Cs = 
 da i ‘ da’ 
 
 Consequently the co-ordinates of the points in which two 
 consecutive characteristics intersect, satisfy the three equations 
 
 du d?u 
 u =,0, wa () - =0; 
 
 + &c. = 0. 
 
170 
 
 and from these equations we may either determine a, y, 
 and x in terms of a, that is, the co-ordinates of the point 
 in which a given characteristic is intersected by the con- 
 secutive; or eliminate a between them, and so find the two 
 equations in w, y, and x, to the curve formed by all the 
 successive points of intersection. This curve will be touched 
 by all the characteristics, and will be the edge of regression 
 of the envelope; for, as explained in the case of develop- 
 able surfaces, the portions of the characteristics which are 
 on contrary sides of their points of contact with the curve 
 under consideration, will form two distinct sheets of the 
 envelope, and these sheets will touch one another in that 
 curve and have it for their common limit. 
 
 195. The propriety of the term Characteristic will be 
 more apparent, if we generalise the equation w= 0, and 
 suppose it to contain together with a a function of a, 
 @ (a), so as to have the form 
 
 f 2, Y, ®, ay @ (a)} =n) 
 
 Then whatever form we give to @, the intersection of two 
 consecutive surfaces represented by this equation will always 
 be a curve of the same sort, and will therefore offer a 
 character common to the whole family of surfaces. For this 
 curve of intersection will be represented by the equations 
 
 Smaking ¢@ (a) = B} 
 Bes df dB 
 Tid, ys. a, (3) = 0; at, da Saas 
 
 and we see that by changing A we shall only alter the 
 constant parts of these equations, without at all affecting 
 the way in which a, y, # enter into them; so that for all 
 values of B; the curve represented by them will be of the 
 same species. 
 
 196. ‘To determine the equation to the envelope when 
 the equation to the series of surfaces contains two independent 
 parameters. 
 
 Suppose the equation #«=0 to contain two parameters 
 
171 
 
 a and £ independent of one another; then the equation to 
 the surface which differs insensibly from that represented by 
 u = 0, will be 
 
 du 
 1 ——~ fa Sh. OB + &e. me = 
 da 
 
 and the co-ordinates of the cy in which these surfaces 
 intersect must satisfy their two equations, that is, they must 
 satisfy 
 
 a 
 wu = 0, ot om 0B + &c. =0; 
 
 or if the surfaces be consecutive 
 
 - du x du 
 
 Radia aay. 
 
 where A = limit of d8+da; but since A may have any value, 
 the latter equation resolves itself into 
 
 du du 
 da ” dB 
 these then together with w=0, are the three equations for 
 determining the curve of intersection of two consecutive sur- 
 faces; from whence we may obtain its two equations involving 
 only one of the parameters; and finally by eliminating that 
 parameter we may obtain the equation to the surface generated 
 by the perpetual intersection of the first series of surfaces. 
 
 = 0, 
 
SECTION IX. 
 
 ON THE CURVATURES OF CURVES IN SPACE. 
 
 197. Preparatory to finding the radius of curvature, 
 and evolutes of a curve in space, consider figure (59), where 
 for the curve is substituted an equilateral polygon mm'‘m”..., 
 and through the middle points of its sides are drawn planes 
 respectively perpendicular to them, which intersect, two and 
 two, in the lines-kh, k’h’, k”h”, &c. Then the plane which 
 contains the two consecutive sides mm’, m’m”, is perpendicular 
 to each of the planes gh, g’h’, and therefore to their common 
 intersection kh; let kh meet this plane in the point q, then 
 q is the center of a circle passing through the three angles 
 m, m’,m”; and every point in the line kh is likewise equi- 
 distant from the same three angles. 
 
 The lines kh, kh’, &c. will be parallel only when the 
 sides of the polygon mm’m”...are in the same plane; in other 
 cases, if they be produced till each meets its consecutive, they 
 will form a polygon hop..., the angular points of which are 
 equidistant from four consecutive angles of the first polygon 
 mm 'm'’...The point o for instance, since it is situated in kA, is 
 equidistant from m, m’, m”; and again, being situated in k'h’, 
 it is equidistant from m’, m”, m’”; that is, it is the center of a 
 
 wr 
 
 sphere passing through four consecutive angles, m, m’, m”, m’”. 
 
 198. The preceding results being true when the number 
 of sides of the polygon is indefinitely increased, it follows that 
 the normal planes of a curve generate, by their perpetual 
 intersection, a curve surface; also, since of the lines of 
 intersection kh, kh’, &c. every two consecutive ones are in the 
 same plane, the surface which they generate is developable. 
 
173 
 
 199. On the same supposition, the polygon hop...be- 
 comes a curve, to which all the lines ko, k’p, &c. are tangents ; 
 that is, hop...is constantly touched by the straight line 
 which generates the developable surface, and is therefore the 
 limit to the surface, for no part of the surface can fall within 
 the space towards which the curve is concave. Moreover its 
 tangent, being produced both ways from the point of contact, 
 generates two sheets of the surface which are united and 
 terminated in the curve; consequently, as was explained in 
 Art. (181), the curve hop...is called the edge of regression 
 of the developable surface, from the analogy it bears to a 
 point of regression in a plane curve. This curve is also the 
 locus of the points of intersection of three consecutive normal 
 planes, or of the centers of spheres passing through four 
 consecutive points of the first curve mm’m”..., that is, it is 
 the locus of the center of spherical curvature. 
 
 200. The plane ggg’ passing through two consecutive 
 tangents to the curve, and which is perpendicular to the 
 intersection of two consecutive normal planes, is the osculating 
 plane, whose equation was found in Art. 127; and the point 
 qg, in which that intersection meets it, is called the center of 
 
 absolute curvature. 
 
 201. To find the equation to the surface generated by 
 the perpetual intersections of the normal planes to a curve 
 in space. 
 
 Let y= («), = = y (2), be ‘the equations to the curve, 
 
 dy dy 
 
 and let qn? da? &c. be denoted by 9, y”, &c.; then we have 
 
 seen Art. (130), that the equation to the normal plane 
 X—a2+(¥-y)y +(Z-%)% =0, (1) 
 and the derived equation with respect to a, viz. 
 (Y—-y)y + (Z—-s) a" -1-y*-x°=0, (2); 
 
 are the equations to the line of intersection of two consecu- 
 tive normal planes; XY, Y, Z, denoting the co-ordinates 
 
174 
 
 of any point in that intersection. If therefore we eliminate 
 w between them, we shall have the equation to the surface 
 generated by the line of intersection in all its successive 
 positions. On this surface lie all the evolutes of the curve, 
 as will be shewn; it will be a cylinder if the given curve lie 
 all in one plane, and if the given curve be traced on the 
 surface of a sphere it will be a cone with its vertex in the 
 center of the sphere. 
 
 202. ‘To find the radius, and co-ordinates of the center, 
 of spherical curvature of a curve of double curvature at any 
 point. 
 
 The center of spherical curvature, or of the sphere which 
 passes through four consecutive points of a curve, is the point 
 of intersection of two consecutive generating lines of the sur- 
 face considered in the last article; hence we must join to 
 equations (1) and (2) of last article, the derived equations 
 with respect to v, as explained Art. (185). 
 
 If therefore , Y, Z be the co-ordinates of the center 
 of spherical curvature, we have in order to determine them, 
 the equations, 
 
 X—-a+(Y-y)y + (Z—2)2 =0, 
 (Y-y)y" + (Z—-s)2"- A +y? + 2%) =0, 
 (P-y)y" + (Z—#)2" — 3 (y'y" + 2'2") = 0, 
 and the radius of the sphere 
 - a V/(X— a) + (Fy) + (Z- 2)". 
 
 Moreover if from these three equations we eliminate a, 
 there will result two equations between X, Y, Z, which will 
 be those to the locus of the center of spherical curvature, or 
 to the edge of regression of the developable surface generated 
 by the intersections of the normal planes. 
 
 203. To find the radius, and co-ordinates of the center 
 of the circle of absolute curvature. 
 
175 
 
 The center of absolute curvature is the point where the 
 line of intersection of two consecutive normal planes meets the 
 osculating plane. 
 
 Let s be the length of a curve from a given point to a 
 point 7, y, x; and instead of taking two equations between 
 &, y, x to represent the curve, let us suppose each of them to 
 be a function of another variable ¢; and let a’, wv’, &c. denote 
 their differential coefficients with respect to ¢; then the equation 
 to the osculating plane at wyz is 
 
 (y's — 2'y") (X - a) 
 + (2 a” —a's")(Y —y) + (e'y" -y'2")(Z-z)=0. (1) 
 The equation to the normal plane at the same point is 
 we (X-—wv) +y (Y-y) +2 (Z-2)=0, (2) 
 and this, in conjunction with the equation obtained by dif- 
 ferentiating it . 
 a (X-a2)+y" (F-y) +2" (Z-2)=8", (3) 
 (since a? + y? +2" =s8", and». a’ a” +y'y" + 32/2" =8's") (4), 
 defines the line of ultimate intersection of two consecutive 
 normal planes (Art. 130). Hence if .Y, Y, Z, satisfy at once 
 
 these three equations, they are the co-ordinates of the center 
 of absolute curvature at wyz; and the radius R will equal 
 
 V(X — «)? + (¥ -y)? + (Z - 8)? 
 
 Now from (1) and (2) taking account of equations (4), we 
 get by ordinary algebraical operations, 
 
 AX —& Y-y Z-— 
 savas’ sy —ys" sls" — 2! 8" 
 5 : from (3) 
 rom (; 
 s a wy + y”? 4 9/2 _ 32 oe! 4. yf” 4. 9/2 — gl ? 
 
 which equations give the radius and co-ordinates of the center 
 of the circle of absolute curvature. 
 
 Cor. If we make w the independent variable by putting 
 
176 
 
 , Ad ”? d a 
 t=, so that 7 =1, 7 =0, and y'sy &c., become et “ y 
 die da’ 
 
 we get by reduction 
 (1+ y? +s”) 
 
 = ja eh Gisele yy 
 Sy + 8? + (2"y! — xy") 
 
 204. Also if we make s, the length of the arc, the in- 
 dependent variable instead of ¢, we get immediately 
 
 1 
 R ad J x se y” “ 2 
 i” dx 
 where wv denotes ds?” &c.; and the expressions for the co- 
 ordinates of the center assume the following simple forms, 
 a ” 
 A-a“7= a” 4/8? 43 1129 Y-y= a? 4 of 4 agit? 
 y gl! 
 -Z%= 
 
 WN 
 
 wey + 3/2 
 
 205. A curve of double curvature may either have three 
 consecutive elements in the same plane, or two consecutive 
 elements in the same straight line; the first is called a simple 
 inflexion, that is, when the curve at that point becomes plane, 
 and therefore the radius of spherical curvature infinite; and 
 the second a double inflexion, as necessarily including the 
 former, that is, when the curve at that point becomes a 
 straight line, and therefore the radius of absolute curvature 
 
 infinite. 
 
 If in Art. 202, we actually determine the values of XY — a, 
 Y-y, Z-—z, we find for their common denominator the 
 
 expression 
 
 vIn OP she Md 2 
 wy -YyY ® 3 
 hence the condition 
 
177 
 
 expressing that the center of spherical curvature is at an 
 infinite distance, will determine the points of simple inflexion, 
 and agrees with Art. 128. 
 
 At points of double inflexion, the radius of absolute curva- 
 ture changes its sign, and the curve, after being concave in one 
 direction, crosses its tangent line and becomes concave in the 
 opposite direction; hence also its projections on the planes of 
 sv and wy cross their tangents, and have corresponding points 
 of contrary flexure; at such points therefore, and at cusps, 
 
 d’ z | d’y 
 
 ——=0, Or ©, — =0, or ©; 
 
 da® div® 
 which are the equations for determining the abscissee; and 
 which we observe make the expression for the radius of absolute 
 curvature infinite or evanescent, as ought to be the case. 
 
 206. The tangent lines of the given curve mm’,..(fig. 59) 
 will generate a developable surface, whose edge of regression is 
 the curve mm’...; and since the tangent line is the intersection 
 of two consecutive osculating planes, this surface would also be 
 generated by the perpetual intersection of the osculating planes. 
 But the lines of intersection of the normal planes are perpen- 
 dicular to the osculating planes; therefore the angle between 
 two consecutive lines of intersection, that is, two consecutive 
 tangents of op..., is equal to the angle between two consecutive 
 osculating planes of mm’... Again, the angle between two 
 consecutive tangents of mm’... is equal to the angle between 
 its corresponding normal planes, that is, to the angle between 
 two consecutive osculating planes of op... If therefore we call 
 the angle between two consecutive tangents of a curve of 
 double curvature its first flewion, and that between two con- 
 secutive osculating planes its second flewion, we may enunciate 
 the above properties as follows. The first flexion of a curve 
 of double curvature is equal to the second flexion cf the edge 
 of regression of the developable surface generated by its 
 normal planes; and the first flexion of the latter, is equal to 
 
 the second flexion of the former. 
 12 
 
178 
 
 Evolutes of a curve in space. 
 
 207. Every curve, whether plane or of double curvature, 
 has an infinite number of evolutes, all of which lie on the 
 developable surface generated by the perpetual intersection of 
 the normal planes. 
 
 Recurring to fig. 59, since kA is perpendicular to the plane 
 through two consecutive sides mm’, m'm”, of the polygon, any 
 point f in it is equidistant from the middle points of the sides 
 g and g’. If therefore we draw gf in the first normal plane, 
 and g’ff’ in the second, the lines gf, g’f are inclined at the 
 same angle to kh; and if with center f and radius fg a circle 
 be described, it will touch the two consecutive sides mm’, m'm” 
 in their middle points g and g’, and when the describing radius 
 comes to g’ it will be confounded with g’f’. In like manner if 
 we draw 2 f'f” in the third normal plane, the lines g’f’, gf’ 
 are equally inclined to k’h’; and a circle described from f’ with 
 radius f’g’ will touch the consecutive sides m’m”, mm” in 
 their middle points, and the describing radius will be con- 
 founded with g”f” at the point g”. By continuing this con- 
 struction the polygon fff’... will be formed, by unwinding 
 a thread gf from which, circular arcs, touching every two 
 consecutive sides in their middle points, will be described. 
 Therefore when the number of sides is indefinitely increased, 
 the polygon fff’... will become a curve traced on the develop- 
 able surface formed by the intersections of the normal planes ; 
 and by unwinding a string from it, the proposed curve of 
 double curvature will be traced out. 
 
 Also since the direction of the initial radius gf is arbitrary, 
 any other direction would have produced a curve endowed 
 with the same properties as ff/f”...; it follows therefore that 
 every curve has an infinite number of evolutes all contained 
 on the surface generated by the perpetual intersections of its 
 normal planes; in the case of a plane curve all the evolutes, 
 except that in its own plane, are curves of double curvature, 
 traced on the right cylinder whose base is the evolute in its 
 own plane. 
 
179 
 
 208. Since ze fk =e" f'k' = h'f'f’, if the plane hk’ 
 were brought into the same plane with h’k” by being turned 
 about kh’, ff’ would be brought into the same straight line 
 with ff’; and since the same may be proved successively of 
 all the other sides of the polygon fff’..., it follows that if 
 the surface formed by the intersections of the normal planes 
 were developed, the evolute ff'f’... would become a straight 
 line, and therefore would be formed by stretching a thread 
 in the direction gf and applying it freely to that surface. 
 Hence if from any point in a curve of double curvature, any 
 line be drawn touching the surface formed by the normal 
 planes, and be applied freely to that surface, it will form an 
 evolute of the proposed curve; and will be the shortest line 
 that can be drawn between any two points of the surface, 
 through which it passes. 
 
 209. It is to be observed that the centers of absolute 
 curvature are not situated on an evolute of the curve. 
 
 For if g’q’ be perpendicular to kh’, then it cannot pass 
 through the point g; because it would then coincide with g"q, 
 and therefore cut kh at right angles, which would require that 
 kh and kh’ should be parallel; and this can never happen as 
 long as the proposed curve is of double curvature. Hence 
 the consecutive radii of absolute curvature gq, gq’, since they 
 are in different planes, and do not meet one another in the 
 common intersection of those planes, therefore they do not 
 meet one another at all; and therefore cannot be consecutive 
 tangents of the same curve; that is, the locus of the center 
 of absolute curvature is not an evolute. 
 
 210. ‘To find the equations to any proposed evolute of 
 a curve of double curvature. 
 
 As we know the equation to the surface on which all the 
 evolutes lie (Art. 201), it only remains to find the equation 
 which particularizes a given evolute. Let X, Y, be the co- 
 ordinates of a point in the projection of an evolute, wv, y, the 
 co-ordinates of the corresponding point in the projection of 
 
 12—2 
 
180 
 
 the curve; then a tangent to the projection of the evolute, 
 must pass through the corresponding point in the projection 
 of the curve; 
 
 Sed Oey 
 
 Ad Xs 
 
 This joined to the two equations 
 oe +4 Yo y) y+ (Z—2)x'=0, 
 (VY —y)y"+(Z —2)2"- (14+ y? +2”) =0, 
 
 which belong to the generating line of the developable surface 
 containing the evolutes, will give by eliminating w the two 
 equations to the evolute, one of them being a differential of 
 the first order; and its integral will introduce an arbitrary 
 constant, by means of which the evolute may be made to 
 satisfy that condition which particularizes it; as, for instance, 
 to pass through a given point of the developable surface on 
 which it is traced. 
 
SECTION X. 
 
 ON THE CURVATURE OF SURFACES. 
 
 211. ‘To find the requisite conditions for a contact of 
 the first, second, &c. order, between two surfaces. 
 
 If two surfaces, referred to the same origin and axes, 
 pass through the same point, the co-ordinates of which are 
 xv, y, %; and if we change w into w+h, and y into y +k, 
 the equation to the first surface will give for the value of 
 the new ordinate, 
 
 st+ph+qk+4 (rh? + 2shk + tk’) + &e. 
 and the equation to the second surface 
 s+ Ph+Qk+h(RW4+2ShK 4 Th) + &e.; 
 
 the distance of the surfaces, measured in the direction of their 
 ordinates, will therefore be expressed by 
 
 (P-p)h+(Q-gQk+3}(R-r)h?+2(S-s)hk+(T-d kh} + &e. 
 
 If we suppose the equation to the second surface to 
 contain a certain number of arbitrary constants, we may 
 determine them so as to make the first terms of this dis- 
 tance vanish; and it will follow that any other surface, for 
 which these terms do not disappear, cannot be situated 
 between the two former with reference to the points which 
 are contiguous to their common point; at least so long as 
 we take h and & so small, that the sum of the terms of 
 the first order may be more considerable than that of all 
 the terms of succeeding orders. When we have P—p=0, 
 Q - q=0, the surfaces will have a contact of the first order ; 
 if besides these, we have R-r=0, S—s=0, T'—-?#=0, 
 the contact -will be of the second order, and so on. 
 
182 
 
 212. Let therefore V=0 be an equation between three 
 variables a’, y’, x and a certain number of arbitrary constants, 
 and according to the usual notation, let P, Q, R, &c. denote 
 the partial differential coefficients of ’ with respect to a’ and 
 y 3 by the determination of the constants we may make this 
 surface have with one completely given (the co-ordinates of 
 which we shall represent by a, y, x, and the partial differential 
 coefficients of z, by p, qg, &c.) a contact of an order which will 
 depend on their number. ‘The first condition to be satisfied is 
 that the surfaces may have a common point, that is, that by 
 changing #’ into a, and y’ into y, in V= 0, we may find x’ = x. 
 Besides this, for a contact of the first order, upon making the 
 same substitutions for # and y’, we must have P= p, Q=@q;3 
 and for one of the second, besides all the foregoing conditions, 
 we must have R=r7r, S=s, J'= 
 
 Hence it appears that a contact of the first order requires 
 three arbitrary constants, and one of the second, six; and 
 in order to have a contact of the » order, since the terms 
 of three, four, &c. dimensions in # and & in the develop- 
 ment of the difference of the ordinates in the preceding 
 Article are, in number, 4, 5, &c., the number of disposable . 
 constants must be | 
 
 14+424+3+4&c.4+ (n+ 1)=4$(n +1) (7 + 2). 
 
 213. But if we confine our attention to the sections 
 of the surfaces made by a plane containing their common 
 ordinate #, and inclined to the plane of za at an angle 
 whose tangent is m, then, making k=mh, the difference 
 of the consecutive ordinates in these sections will be 
 
 $P—p+(Q-9)mth+4{R-r+2(S—s)m+(T-t)m t{th’+&e. 
 Hence if the surfaces have a common point and a common 
 tangent plane at that point, that is, if upon making a’=a, 
 and y= y, we have x=, P=p, Q=gq, then the first term 
 
 of the above difference will disappear, and the second will 
 also vanish if we add a fourth condition, viz. 
 
 R-7r+2(S-s)m+(T'-t)m =0.— 
 
183 
 
 Hence four disposable constants in the equation to a surface 
 are sufficient in order to establish between it and one com- 
 pletely given a contact of the first order at a proposed point, 
 and also to establish a contact of the second order between 
 the sections of the surfaces made by a See: plane containing 
 their common ordinate. 
 
 214. Previously to applying these considerations to es- 
 timate the curvature of any surface at a proposed point, we 
 may observe that we cannot attempt to assimilate the curvature 
 of a surface in general to that of a sphere; because in the 
 latter the curvature is uniform about the same normal, whereas 
 for surfaces in general that is far from being the case. The 
 mode of proceeding must therefore be to imagine several planes 
 drawn through the normal at the point under consideration, 
 to calculate the radius of curvature of each of these sections 
 at that point, and by comparing them to judge of the greater 
 or smaller curvature of the surface in those directions about 
 the point, as well as towards what parts the curvature is 
 turned; for, as we know, certain surfaces contiguous to any 
 point are situated partly above and partly below the tangent 
 plane at that point. 
 
 In making different planes pass through the same point 
 of a surface, so that some contain the normal to the surface 
 at that point, and some do not, we shall find the radii of cur- 
 vature relative to that point both of the normal and oblique 
 sections, bearing remarkable relations to one another, inde- 
 pendent of the particular form of the surface. We shall 
 begin by demonstrating the simple relation which exists be- 
 tween the radii of curvature of a normal and an oblique 
 section, made by planes passing through the same tangent 
 line to a surface, first noticed by Meunier. We shall then, 
 by the doctrine of contacts, deduce the radius of curvature 
 of any normal section of a surface; and afterwards prove 
 Euler’s theorems relative to the curvatures of the principal 
 normal sections of a surface. ‘The circumstances under which 
 the normal at any point may be intersected by the normals 
 at immediately adjacent points will next fall under considera- 
 
184 
 
 tion, as throwing great light on the subject and leading to 
 many results of interest. 
 
 215. If a normal and an oblique section of a surface 
 be made by planes passing through the same tangent line 
 to the surface, the radius of curvature of the oblique section 
 is equal to the projection on its plane, of the radius of 
 curvature of the normal: section. 
 
 Let the tangent plane to the surface be the plane of 
 wy, the point of contact the origin, and the tangent line the 
 axis of w; then the normal to the surface will be the axis 
 of =; let OP (fig. 62) be the normal section in the plane of 
 sv, OP the oblique section made by a plane s’Ow through 
 Ow, and inclined to the normal section at an 22Oz=6@; 
 NP, NF’, ordinates to the two curves corresponding to the 
 common abscissa ON =h; also let NP=2, and let h, &k, 2’ 
 be co-ordinates of P’, Then, (assuming that in any plane 
 curve when the axis of the abscisse is a normal at the origin, 
 the radius of curvature at the origin is equal to 4 limit of 
 
 oe eee) 
 ———— | we.have 
 abscissa 
 milidevorMbureature of Onests0 oe ion: cee 
 FA=radius of curvature o at O= 5 Imit o WP? 
 ON? 
 
 Casita ; ; a ( ee ise ye sa ee... 
 R= radius of curvature of OP at O=4% limit of NP” 
 
 ~ 
 
 NP’ x’ sec 
 — = limit of ——= limit of - -— =sec@. limit of 
 NP x 
 
 a | & 
 
 But since the plane of wy is the tangent plane at O, 
 p=0, q=C, and s=trh?+shk+4tkh + &e. 
 
 and making k = 0, 
 
 ee 
 pas trh? +t+—.h' + &e.3 
 
185 
 
 tk k : ; 
 because limit of re since Ow is a tangent to the pro- 
 ) 
 
 jection of OP’ on the plane of wy; 
 
 . R= Rcoos8. 
 
 Cor. Hence it follows that the osculating circles of all 
 sections of a curve surface that have a common tangent 
 line, are situated on the surface of a sphere the radius of 
 which is the radius of curvature of the normal section pass- 
 ing through the same tangent line; for OA’= OA cos AOA 
 (and consequently AdA’O is a right angle) if OA be the 
 diameter of the circle of curvature of the normal, and OA’ 
 that of the circle of curvature of the oblique section; the 
 latter circle is therefore situated on a sphere whose diameter 
 
 is AO. 
 
 216. To find the radius of curvature of any normal 
 section of a surface at a given point, in terms of the co- 
 ordinates of that point. 
 
 Let w, y, x be the co-ordinates of the given point P 
 (fig. 63) of the surface, p, q, 7, &c. the partial differential 
 coefficients of x expressed in terms of those co-ordinates ; 
 PT the tangent line through which the normal section is 
 to pass, which, since it lies in the tangent plane at P, 
 will be determined by the equation to its projection on the 
 plane of vy, viz. y—y=m(a'—«). Then if we determine 
 a sphere having a contact of the first order with the surface 
 at P, and whose section by the vertical plane PQT' has a 
 contact of the second order with the section of the surface 
 by the same plane, all planes passing through P7' will cut 
 the sphere in circles, which are the circles of curvature to 
 the corresponding sections of the surface, and therefore the 
 radius of the sphere will be the radius of curvature of the 
 normal section; and this we are enabled to do, because the 
 four disposable constants in the equation to a sphere will 
 enable us to satisfy the four requisite conditions, which are, 
 
186 
 
 that upon making 2’=«a and y'=y in the equation to the 
 sphere, we must have 
 
 s=2, P=p, Q=q, R-7r+2(S-s)m4+(T-t)m’ =0. 
 
 Now the equation to the sphere gives, by differentiation, 
 (2 — a)? + (y'- By + @'- y= 8 
 a’—-at P(s-y)=0, y'-B+ Q(2’-y) =9, 
 14+-P°+R(s'-y)=0, PQ+S8(%-y)=0, 14Q°?+T7(s'-y)=0; 
 
 hence, there result between the constants a, B, yy, 0, and 
 the co-ordinates vw, y, x, the following relations 
 
 (w—a)’+(y- 8)? + @-yyP =e, 
 w-atp(s-y)=0, y-B+q(?-y) =9, 
 1 2 2 
 Mica +rgo( Pt 4) m + (= +t) m= ; 
 e—%¥ z-¥ z 
 
 The latter gives 
 
 1+p°+2pqm+(14+q)m 
 y+2sm+tm 
 
 e-y=— 
 
 and then a, 2, 6 are known from the equations 
 w-a=—p(z—-y), y-B=-9g(e-y), P=(e-y)P(+p+e); 
 
 Vi1l+ pit’ {1 +m? + (p+qm)°} 
 
 . the radius = 
 ry +2sm +tm’ 
 
 Hence we have determined the radius and co-ordinates 
 of the center of the circle of curvature of the normal 
 section at a point wy, whose intersection with the tangent 
 plane at that point is projected into a line represented by 
 the equation 
 
 y-—y=m(a'— 2). 
 
 Cor. Suppose the tangent line to make angles dA, pu, v 
 with the axes of a, y, x, and let o be the length of the 
 arc of the normal section, then (Art. 28 and 131), 
 
187 
 
 COS ph 
 m =——, and 
 cosr 
 
 do HAG (54) +m + ( + mg = (2 2 
 (=) poe: da +(= m4 P iy = | , 
 
 hence, by substitution, we get for the radius of curvature 
 
 Y1l+p+¢ 
 
 ~ # (cosA)* + 28 cos dA Cos p + t (cos nu)?” 
 
 217. But if the equation to the surface be proposed 
 under the form F'(a, y, #) = 0, the radius of curvature of a 
 normal section may be more readily obtained by considering 
 (Art. 200) that the center of curvature of the section is the 
 point in which the normal meets the line of intersection of two 
 consecutive normal planes. 
 
 dk d dF 
 lor ee eae eal — = W, so that 
 dx dy dz 
 
 Ude+ Vdy+ Wdz=0; 
 then the equations to the normal at wy are 
 AX-« Y- Z—s 
 a ays a ees Q, suppose (1). 
 
 Now if ds be an element of the arc of the normal section, 
 and da, dy, dz its projections on the co-ordinate axes, the 
 equations to the normal plane to the section will be (Cor. 
 Puric. 120. } 
 
 (X -x)dx+(Y-y)dy+(Z-—sz)dz=0; (2) 
 
 and the line of intersection of this with its consecutive is given 
 by (2), together with its differential, viz. 
 
 (X — 2) @a+(Y-y) dy +(Z—2)a@x=ds*. (3) 
 
 Hence if XY, Y, Z be determined so as to satisfy the above 
 three equations, they will be the co-ordinates of the center of 
 curvature of the normal section; and if R be the radius of 
 curvature, we shall have from (1) 
 
 CMS OL yy + (2-3 x) 
 
 v= \/U? + VV? + W?.Q 
 
188 
 
 but from (3) we get 
 (Uda + Vay + Wd’s) Q= ds’; 
 
 ds?.4/U? 4+ V2.4 W 
 
 Oy are ae 
 Udu+ Vay + Wax 
 
 But since Ude + Vdy + Wdz =0, if we differentiate 
 and call 
 
 dF d’F dF 
 
 Teak —-=UV, ——~=W; 
 
 dye? dx" 
 aon pains aay ihe fatdtin tie 
 
 dydzs’ dads’ dudy 
 
 we get 
 UPx+ Va@y + Wd’: + uda’ + vdy + wdz + 2uwdydz 
 +2vdedz+2wdady =0; 
 consequently, substituting, and calling 
 dx } dy dz 
 
 Bei —=m, —=N, 
 ds ds 
 
 ds 
 
 et 
 
 we have 
 \/ U? + V24 W 
 ~— at mv + niw + 2imw' + 2lnv'+ 2mnw 
 
 3 
 
 which coincides with the former expression for R, if we 
 
 suppose (wv, y, 7) =2—f(a, y) = 0. 
 
 218. Hence we can find the conditions, in order that 
 the curvatures of the normal sections of a surface at a pro- 
 posed point may be all in the same direction, or in opposite 
 directions; that is, in order that the surface may be convex 
 or non-convex about a proposed point. 
 
 If in the value of # (Art. 216) we always take the radical 
 positive, as the numerator 1 + m* + (p+mq)’ is incapable of 
 changing its sign, the sign of J will depend upon that of 
 its denominator 
 
 ] 
 r+ 28m RE in S(mt +s)? +rt— 8°}; 
 
189 
 
 if therefore xv, y, x the co-ordinates of the proposed point be 
 such that rf — s*>0, then # cannot change its sign for any 
 value of m, and the surface contiguous to that point will 
 be situated entirely on thé same side of the tangent plane, 
 or will be convex at that point; and if this be true for 
 every point, as in the ellipsoid, the surface will be entirely 
 convex. If, on the contrary, r¢—s*<0, there are two 
 values of m for which the denominator will vanish, and R 
 will be infinite and will change its sign as m passes through 
 each of these values; therefore the surface will have opposite 
 curvatures about the point in question, or will be non-convex. 
 Thus in the surfaces considered (Art. 176), of which the 
 equation is 
 
 gr—2pqs+pt=0, or (qr—ps)’ +p? (rt —s’) =0, 
 
 rt —s° is necessarily negative, and therefore surfaces of this 
 class are non-convex at every point. 
 
 For developable surfaces, where r¢— s* = 0, the denomi- 
 nator of # is a perfect square, and consequently retains the 
 same sign; therefore the radius of curvature will haye the 
 same sign for every point of the surface; only it will at every 
 point be infinite when mt +s = 0, the direction so determined 
 evidently coinciding with that of the generating line passing 
 through the point. 
 
 219. 'To determine the normal sections of greatest and 
 least curvature at a given point of a curve surface. 
 
 For the same point of the surface, the expression for R 
 varies with m, and we may find what value of m will make it 
 a maximum or minimum by putting the differential coefficient 
 of & with respect to m (for the quantities p, q, r, &c. are 
 in general independent of the position of the tangent line 
 which fixes the normal section) or rather of x —y on which R 
 depends, equal to 0; hence putting equation (1) (Art. 216), 
 under the form 
 
 (x -—y) (v7 +28m + tm’) +14 p> + 2pqm + (1 + q’) m® = 0, 
 we have (x -—-y) (s+ tm) + pat (1+q’)m=0; 
 
190 
 
 therefore eliminating successively x —-y and m from these 
 equations, we get 
 
 S(14q°)s—pqt}m?+ }(1+q°)r-(1+p*)t} m— { (14+ p*)s—paqr} =0, 
 (rt-s°)(x-y)?+{(1+p’)t-2pqs+(1 +7 )r}(s-y)+14+p°+q=0. 
 
 These equations being of the second degree, it follows 
 that in general the radius of curvature of the normal section, 
 as the cutting plane revolves about the normal, will have only 
 one maximum and one minimum value; which, in absolute 
 magnitude, may be two maxima or two minima, when the 
 curvature of the surface changes its sign about the given 
 point. These two normal sections, one corresponding to the 
 least, and the other to the greatest radius of curvature, are 
 called the principal sections of the surface relative to the 
 point at which the normal is drawn, and the corresponding 
 radii the principal radii of curvature. 
 
 Cor. Hence the former of the above equations enables 
 us, for each point of the surface, to determine the directions 
 of the principal sections, and we might also shew from it 
 that these directions are at right angles to one another. 
 To determine however the angle between the principal sec- 
 tions in a simple manner, let us suppose the plane of vy to 
 be coincident with the tangent plane at P, or only parallel 
 to it, (this of course we may do without at all altering the 
 form of the surface or the positions of the tangents to the 
 principal sections) then we must put p=0, g=0, and the 
 equation becomes 
 
 Brg ae PGR TTD 28, 
 
 the two roots of which are always possible and satisfy the 
 condition 
 
 therefore the principal sections always exist and are at right 
 
191 
 
 angles to one another. Also the same supposition gives for 
 the radius of curvature of any normal section (Art. 216), 
 
 1 +m? 
 e+ 25m4+tm- 
 
 Having however made the foregoing process answer its pur- 
 pose of giving the radius of curvature of any normal section 
 of a surface in terms of the co-ordinates of the point, we shall 
 not deduce from it all the important results which it is capable 
 of furnishing, but obtain them by a direct and much simpler 
 method. 
 
 220. ‘The sum of the curvatures of any two normal sec- 
 tions at right angles to one another, is a constant quantity at 
 the same point of a surface. 
 
 Let AO, the normal at any point A of a curve surface, 
 be the axis of zs, and wdAy, the tangent plane at the same 
 point, the plane of wy (fig. 64). Let AB be a section of the 
 surface made by any plane passing through AO, AC the 
 corresponding section of the tangent plane, which therefore 
 touches the plane curve 4B at 4; and let y = tan@.« be the 
 equation to AC, so that 0@=ZNAC, Then if 40 =R be the 
 radius of curvature of the plane curve 4B at A, and BC be 
 parallel to 40, we have by a well-known theorem 
 
 A 2 
 R= 3 limit of = - 
 
 Let dN=h, NC =k, be the co-ordinates of C; 
 “ BCa=trh?+shk+ttkh + &e, 
 
 for p=0, gq =0, since the plane of wy is the tangent plane ; 
 also kA = tan@0.h; 
 
 +i 
 * R= limit of —— —_ —__— 
 2 Lrht-+shk+ 4th + &e. 
 a): Petal Oe 
 = limit of ; 
 
 dr 
 r+2stan@+étan?@ + met htan?@ + &e, 
 aw 
 
192 
 
 1 + tan?@ 
 
 or 2 = — 
 r+ 2stan0+¢tan’@’ 
 
 where 7, s, £ are the values of 
 
 d’z d’x d’s 
 du?’ dady’ dy’ 
 
 at the point A derived from the equation to the surface. 
 Hence, if R’ be the radius of curvature of another section 
 inclined at an angle = 90°+ @ to AN, we have 
 
 1 : Ke 
 rea 7 cos’ @ + 2s cos @ sin @ + ¢ sin’ 0, 
 
 ] ; : . 
 —=rsin?@ — 2ssin 0 cos@ + ¢ cos’ @; 
 
 R' 
 
 which (taking, as usual, the inverse of the radius of curva- 
 ture to measure the curvature) expresses that the sum of 
 the curvatures of any two normal sections at right angles 
 to one another, is a constant quantity at the same point of 
 the surface. 
 
 221. Of all sections of a curve surface made by planes 
 drawn through the normal at any point, to determine those of 
 greatest and least curvature at that point; and to shew that 
 they are at right angles to one another. 
 
 We have 
 fay dart ; 
 R= 7 cos’ O + 28 sinO cos O + ¢ sin’ A, 
 he r (1 + cos 20) + 28 sin 20 + ¢(1 — cos 20) 
 
 =r+i#+4 (r —- ¢t)cos20 + 28 sin 20, 
 
 which is to be a maximum or minimum by the variation of 0; 
 
193 
 
 therefore putting its differential coefficient with respect to @ 
 equal to zero, we get 
 
 — (r — f) sin20 + 2s cos20 =0; 
 
 28 
 *, tan20 = ari tan (180° +20); (1) 
 
 , 28 yr—t 
 sin 20 = £ ——______—. , cos 20 = ok Sey 
 J (r — ty + 48° VJ (r —t)? + 48 
 2 (r — ty 48° 
 We —_ = 7 + Zt + yg de a ; 
 a JS (r—t +48? SA (r—f) + 48 
 
 : 1 17 Sete ot 
 or Fy es (r — t)* + 48°, 
 
 which are the values of the reciprocals of the greatest and 
 least radius of curvature; and the positions of the sections 
 to which they belong are fixed by the angles 0 and 90° + @ 
 obtained from equation (1); consequently the sections of 
 greatest and least curvature at any point of a surface, called 
 the principal sections at that point, are at right angles to 
 one another. 
 
 222. The curvature of any normal section is equal to 
 the sum of the curvatures of the principal sections, multiplied 
 respectively by the squares of the cosines of the angles which 
 its plane forms with their planes. 
 
 Suppose the axes of w and y to be drawn in the principal 
 28 
 
 planes, then since 0=0 and tan26 = , we must have 
 
 yr —t 
 s = 0; hence 
 1 : 
 2 cet 
 —=rcos'@ + ¢sin’@; 
 R 
 consequently, if we denote by p and p’ the radii of curva- 
 ture of the principal sections corresponding to @=0 and 
 8 = 90°, we have 
 
 1 
 wi ye 
 p p 
 
194 
 
 therefore substituting, we have for any normal section, 
 
 hp roe D daisy 
 a p 
 
 This theorem, due to Euler, joined to that of Meunier, 
 contains the whole theory of the curvature of surfaces; for 
 it is hence sufficient to know at any point of a surface the 
 directions and curvatures of the principal sections, to deduce 
 the curvature at that point of every other section, normal or 
 
 oblique. 
 
 223. When the two principal radii have the same sign, 
 the formula 
 ha —cos’@ + oa sin® 0 
 La p 
 proves that every normal section will have a radius of that 
 sign, and therefore the surface will be entirely on the same 
 side of the tangent plane at the proposed point; also the 
 principal radii will be the maximum and minimum values 
 of the radius of curvature at that point, and the form of 
 the surface will be that represented in fig. 65, where AO, 
 AO’ are the principal radii, and AQ is the radius of the 
 intermediate section AP. 
 
 When the principal radii are equal, as well as of the 
 same sign, the formula gives R =p, and therefore all the 
 normal sections have the same curvature, and may be 
 regarded as principal sections; of this we have examples 
 at the vertex of a paraboloid of revolution, and the poles 
 of a spheroid. 
 
 224. When the principal radii are of contrary signs, 
 p positive and p’ negative for instance, 
 
 1 
 
 1 e € 
 cos® 9 — — sin’ 0, 
 
 it 
 ft aep p 
 
 which vanishes, or R is infinite, when @ has such a value w 
 
195 
 
 that tanw = + Je. hence for values of 6 between 0 and + 
 P 
 
 R is positive, or the section situated above the tangent 
 plane; and for all other values, R is negative, or the section 
 situated below the tangent plane; also p will be the minimum 
 of the positive radii, and p’, numerically, the minimum 
 of the negative radii. Hence at the proposed point, the 
 form of the surface will be that represented in fig. 66; DBE 
 being the intersection of the surface with a sphere center 4; 
 AE, AF sections having respectively a positive and negative 
 radius, and AB the section made by the limiting normal plane 
 
 throu gh AO. 
 
 Thus in the hyperboloid of one sheet, we know that 
 the tangent plane at any point cuts the surface in two straight 
 lines, limiting the normal planes which give positive and 
 negative radii of curvature, or which separate the convex and 
 concave parts of the surface. In surfaces of a higher order, 
 the limiting normal planes will cut them in curves having 
 with the corresponding sections of the tangent planes a contact 
 at least of the second order, since the curvatures of those 
 sections are infinite. In developable surfaces, if we take a 
 principal plane at any point for that of sa”, then s =0, 
 and since rf —s°=0, r=0 or ¢=0; therefore one of the 
 principal radii is infinite. In fact, we know that the tangent 
 plane to a developable surface does not cut it, but touches 
 it along a generating line; and that this generating line is 
 a principal section whose radius of curvature is a maximum 
 and infinite, whilst the minimum radius belongs to the section 
 made perpendicular to the generating line. Hence, the 
 curvature of any surface at every one of its points may be 
 assimilated to that of a surface of the second order at one 
 of its vertices, as will be seen in the following Article, 
 where we shall take a paraboloid, although an ellipsoid would 
 
 do equally well. 
 
 225. To determine a paraboloid of the second order, 
 which shall have at its vertex, a complete contact of the 
 second order with a given surface at a proposed point. 
 
 13—2 
 
196 
 
 Let A (figs. 65 and 66) be the given point of the sur- 
 face, Az the normal, and let the principal sections meet 
 the tangent plane in Aw, Ay, which lines take for the axes. 
 Also, let 
 
 z= os A se 
 2p 2° 
 
 be the equation to a paraboloid, p, p being the principal 
 radii of curvature of the proposed surface at 4. Then if AP 
 be a section of this paraboloid through its axis inclined at 
 an Z2@to ga, and AN=r, PN =2, be co-ordinates of P, 
 the equation to AP, putting rcos@ for # and rsin@ for y, 
 (Art. 101) is 
 
 1 Lhe 
 sat ef cos’ @ + — sin? af : 
 2 if 
 
 which represents a parabola, the reciprocal of whose semi- 
 latus rectum, and therefore of the radius of curvature at 
 
 its vertex, 
 
 Hence the curvature of every section of the paraboloid is 
 the same as that of the corresponding section of the surface, 
 and therefore the paraboloid has a complete contact of the 
 second order with the surface. 
 
 Cor. What we here assume, viz. that if two surfaces, 
 having a common point and common normal at that point, 
 have the curvatures of all normal sections equal, (or, which 
 comes to the same thing, have their principal sections coin- 
 cident and equally curved,) they have a complete contact of 
 the second order at that point, agrees with Art. (211); for 
 this, from the general expression for the radius of curvature 
 of any normal section, requires that x and its differential co- 
 efficients p, q, 7, s, #, should have equal values in the 
 equations to the two surfaces, when in them we substitute 
 the co-ordinates w and y of the point under consideration. 
 
197 
 
 226. In the equation to the paraboloid, suppose x 
 constant, and =e, therefore 
 x y 
 rel oge 
 2 pec 2p ec 
 
 1; 
 
 this is the equation to a section of the paraboloid, or to a 
 section of the surface if ¢ be indefinitely small, Hence it 
 appears that if the surface be cut by a plane parallel to 
 the tangent plane at any point, and indefinitely near to it, 
 the section is ultimately a curve of the second order whose 
 center is in the normal, and axes in the planes of greatest 
 and least curvature; also the square of the diameter in which 
 the curve is intersected by any plane drawn through the normal, 
 is proportional to the radius of curvature of the corresponding 
 section of the surface. 
 
 This curve has been called by Dupin, the Jndicatrix 
 of the surface, because it indicates the directions of the 
 curvatures; if at any point it be an ellipse, p and p’ must 
 have the same sign, that is, the curvatures are in the same 
 direction; if a hyperbola, the curvatures. are in opposite 
 directions; and if a circle, p= p> and the curvature of 
 every normal section is the same. For instance, it has been 
 shewn (Art. 104) that an ellipsoid may be generated in two 
 ways, by a circle of variable radius moving parallel to itself; 
 consequently there are four points on its surface at which 
 a plane, parallel and indefinitely near to the tangent plane, 
 cuts it in a circle; that is, the indicatrix at those points 
 is a circle, and therefore the two radii of curvature equal. 
 These points are called wmbilicit, and are symmetrically 
 placed in the four angles of the principal section containing 
 the greatest and least axes. It is manifest that at these 
 points, a sphere can have a complete contact of the second 
 order with the surface. 
 
 Intersection of Consecutive Normals of a Surface. 
 
 227. As two lines in space will not intersect unless 
 the constants which enter into their equations satisfy a 
 
198 
 
 certain equation of condition, so if from any point in a 
 curve surface at which we have drawn a normal we _ pass 
 to a contiguous point, the normal at the latter point will 
 not intersect that at the former, however near the points 
 are to one another, unless the second point be taken in such 
 a direction as to satisfy the equation of condition for the 
 intersection of the normals. ‘The consideration of consecu- 
 tive normals throws great light on the subject of the 
 curvature of surfaces, as will appear from the following 
 propositions. 
 
 228. Having given a point on a curve surface, to find 
 the directions in which we must pass to consecutive points, 
 in order that the corresponding normals may intersect. 
 
 Take the normal at the given point for the axis of x, 
 and the tangent plane for that of wy, and let h, k, Jd be 
 co-ordinates of a contiguous point situated in a_ section 
 through the normal made by a plane inclined at an angle 
 0 to that of zw, so that K=hAtan@; then the equations to 
 the normal at that point are 
 
 wv —h+p,(2' -l) =0, y —kign(s-J) =0, 
 where p,, g, denote certain functions of h and k (p and q 
 
 d d: 
 denoting the values of 7. and a 
 v y 
 
 fore each = 0). Then in order that this line may meet the 
 axis of x, its two equations must agree in giving the same 
 value for x’ when wv’ and y’ = 0, that is, 
 
 at the origin, and there- 
 
 / 
 
 2 aay 
 s =—+land vs = —+], 
 Pi Nh 
 
 must be the same. But if the points be consecutive, then 
 the ultimate values of x’ must be the same when h = 0, that is, 
 (expanding p, and q,), 
 
 ie A h 1 
 SY scilimit of pee cae 3 
 rh+sk+&c. r+stan@ 
 
199 
 
 k tan @ 
 sh+tk+&c. 8s +#tan9 
 d’°s dz d? x 
 
 (7,8, ¢ being the values of qa®? bray ; ia the origin) must 
 
 and x = limit of 
 
 be identical ; 
 ..s+é#tan@ =rtan@ +s tan’ 0, 
 
 2 tan @ 28 
 
 or tan 29 = ——————. = ——_ 
 1—tan®@ r-—f# 
 
 r! 
 
 an equation which being the same as that for finding the 
 positions of the principal sections through the origin, shews 
 that there are two, and only two, directions at right angles 
 to one another in which we may pass from a proposed point 
 in a curve surface to a consecutive point, so that the normals 
 may intersect; and that these directions coincide with the 
 principal sections through the proposed point. Also the 
 value of 
 
 1 | ee poe oe eee 
 sar+stanO =} frt te (1-2) + 48°} 
 
 shews (Art. 221) that the points where the normal is in- 
 tersected by consecutive normals, are the centers of curvature 
 of the principal sections. 
 
 229. This property of consecutive normals enables us 
 in some cases to determine at once the principal sections of a 
 surface at a proposed point; for instance, in surfaces of 
 revolution the plane of the generating cwrve, supposed to be a 
 plane curve, is necessarily one of those sections, because in 
 it consecutive normals intersect, and therefore a plane through 
 the normal perpendicular to it is the other. The first series 
 of consecutive normals intersect in the evolute of the generating 
 curve, the second in the axis; hence, in a surface of revolution, 
 the loci of the intersections of consecutive normals, or of the 
 centers of curvature of the principal sections, are the axis of 
 revolution and the surface formed by the revolution about 
 that axis of the evolute of the generating curve. Hence 
 
200 
 
 we can find immediately the radius of curvature of any 
 section of a surface of revolution, the principal radii being 
 the radius of curvature of the meridian, and the portion of 
 the normal intercepted between the proposed point and the 
 axis of revolution. 
 
 Lines of Curvature. 
 
 230. A series of points on a surface determined by 
 the condition that the normals at any two consecutive ones 
 intersect is called a line of curvature of the surface. Every 
 point of a surface, as appears from the above investigation, 
 is situated on two curves of this kind, whose dircctions 
 coincide with those of the principal sections through that 
 point, and cut one another at right angles. Thus in start- 
 ing from a point P of a surface (fig. 68), there are two 
 contiguous points P’ and Q the normals at which will inter- 
 sect the normal at P; and if of these we take only that 
 which is in the same direction with KP, namely P’, and 
 then advance in the same direction to the contiguous point 
 whose normal intersects the normal at P’, and so on con- 
 tinually, we shall obtain KPP’n a first line of curvature 
 of the surface; the second line of curvature which passes 
 through P will be obtained in the same manner, and will 
 be RPQE; and as these constructions may be repeated for 
 every point of the surface, we shall thus form two series 
 of lines of curvature dividing the surface into curvilinear 
 quadrilaterals whose sides in space cut one another at right 
 angles. 
 
 It must be observed that the lines of curvature which 
 pass through any point and the principal sections through 
 the same point, although they have common tangent lines 
 at that point, will not usually coincide. Thus in a surface 
 of revolution the meridian PB (fig. 67) and the section 
 through the normal perpendicular to the meridian QPG are 
 the principal sections at P; but only one of them is a line 
 of curvature through P, viz. the meridian; the other line 
 of curvature being manifestly the parallel PD, which has a 
 common tangent with PQ at P. 
 
201 
 
 231. To find the differential equation to the projection 
 of the lines of curvature which pass through a given point of 
 a curve surface. 
 
 The equations to the normal at a point wysx of a curve 
 surface, are 
 
 we —ax@t+ p(s -x)=0, y —-y+q(% — 2) =0; 
 
 and at a contiguous point v+h,y+hkh, x +1, the equations 
 to the normal, are 
 
 wv —xvx—-h+p,(2’-x-l=0, y-y-kig(s'-x-l=0; 
 
 Pir Qs» denoting the same functions of «+h, y +k, that 
 p and q are of w and y, as derived from the given equation 
 to the surface x = f(v, y). If these two normals intersect, 
 their equations must be satisfied by the same values of a’, y’, 2’, 
 which will be the co-ordinates of their point of intersection ; 
 eliminating therefore wv’, y’, x’, between the four preceding 
 equations, we shall have an equation of condition expressing 
 that the normals intersect, and which will establish a relation 
 between A and k& or fix the direction in which we must pass 
 from the proposed point to a consecutive point in order that 
 the normals may intersect. Subtracting the first equations 
 from the second, we have 
 
 —h+(p,— p) (# - 2)-pil=0, -—k + (4-9) (# - 2) -Ql=9, 
 
 and these equations must agree in giving the same value 
 for 2, 
 
 oS h+pl_ pi-P 
  k+ ql ara 
 
 Now let the points be consecutive; then expanding p,, q,, and 
 /, and retaining in their developments only those terms which 
 involve the simple powers of h and k, we must have 
 h + (; *h +sk)(ph k ie A h+sk 
 limit of ec ei) (2 oe ) = limit of sath : 
 k+(q + sh+tkh) (ph + qk) sh + tk 
 But if 6 be the angle which the tangent to the projection of a 
 line of curvature on the plane of wy makes with the axis of a, 
 
202 
 
 we have & = Atan@ ultimately ; hence substituting this in the 
 above equation, and making / vanish, we have 
 1+ p* + pq tand r+stan@ 
 tan0d+pqt+q'tanO s+é#tand@ 
 
 But if we now consider wv and y to be the co-ordinates of the 
 ae : d 
 projection of the line of curvature, we have tan @ = one hence, 
 Lv 
 substituting and reducing, 
 
 {(1 + 9°) s — pqtt (5%) 
 
 airy ’ 
 +f +g) r= (+p) = — {14 ps - par} = 0. 
 
 It remains to substitute for p, q, 7, s, ¢ their values in terms 
 of w and y derived from the equation to the surface and to 
 integrate; the result will be the equation to the projection of 
 the lines of curvature. 
 
 e © . . e ° d ° 
 Since the above equation is of two dimensions in — its 
 x 
 
 integral when completed by the arbitrary constant C’, will be 
 of the form C’ + Cd (a, y) +  (w, y) = 0. Suppose the line 
 of curvature is to pass through a point for which vw =a, y = 6; 
 therefore C? + Cq@ (a, 6) + (a, b) = 0, which will give two 
 values of C; and by substituting them successively in the 
 complete integral, we shall have the equations to the two lines 
 of curvature passing through the given point. 
 
 Cor. We may find the differential equation to the lines 
 of curvature of a surface whose equation is given under the 
 form F' (a, y, ) = 0, as follows. 
 
 Let U, V, W have the same meanings as in Art. 217, then 
 the equations to the normal at a point wyz and at a con- 
 tiguous point are 
 
 X-w Y-y Z-2z 
 _U Ley eye + W) 
 Ape da eae y — dy ee Gass tds 
 Ted Wade? ewe awe 
 
203 
 
 the second of which by expanding and neglecting small quan- 
 tities of the second order becomes 
 
 dv dU # nie = 
 
 Sly be ya ee ms ss 
 
 v + ( v) i +(Y- py wt (Z Aye ye Lares 
 Eliminating 1, Y, Z between (1) and - we get for the 
 condition that consecutive normals may intersect, 
 
 U(dzdV —dydW)+V(dedW —dzdU)+W(dydU-dedV)=0, 
 
 the integral of which joined to the equation F(a, y, x) = 
 will determine the lines of curvature of the surface. If we 
 suppose (x, y, 3) =f(a, y)-#x=0 so that V=p, V=gq, 
 W=-1,dU=rdx+sdy,dV=sdax +tdy, dW =0, we fall 
 upon the equation given above. 
 
 232. The points in which the normal at any point is 
 intersected by the consecutive normals, coincide, as we have 
 seen, with the centers of curvature of the principal sections 
 at that point; and the portions of the normal intercepted 
 between these points and the point of the surface at which 
 it is drawn, (which, as we have seen, are the radii of curvature 
 of the principal sections of the surface at that point,) have been 
 called by Monge the two radii of curvature of the surface at 
 that point, as being the radii of two spheres which alone can 
 touch the surface in two consecutive points. [or if with 
 the points O, O’ (fig. 65) as centers, and radii OA, O'A, we 
 describe spheres, the former will touch the surface along 4G 
 and the latter along AD, because consecutive normals to the 
 surface in both these sections meet 4O. But if with Q, the 
 center of curvature of any section AP, as center, and radius 
 AQ we describe a sphere, it will not touch the surface along 
 AP, because QP cannot be a normal to the surface, since it 
 intersects AO; only, this sphére will have the same tangent 
 plane as the surface at 4, and its section made by the plane 
 AQP will have with the corresponding section of the surface a 
 contact of the second order. 
 
 From the consideration of consecutive normals we can now 
 find the expressions for the lengths of the principal radii of a 
 surface at any point much more easily than by the former 
 method. 
 
204 
 
 233. To determine the radii of curvature at any point 
 of a surface in terms of the co-ordinates of that point, from 
 the consideration of consecutive normals. 
 
 The equations to the normal at a point wyz of a curve 
 surface are 
 
 we —vt+p(s%-xs)=0, y -y+q(s'—-2)=0, 
 
 and at a contiguous point vw +h, y +k, x+/, the equations 
 to the normal are 
 
 wv —v-h+p,(2'-x-l=0, y -y-k+9q,(% —z-/) =09. 
 The values of w’, y’, 2 which simultaneously satisfy these 
 four equations are the co-ordinates of the point of inter- 
 
 section of the normals; subtracting the two former from the 
 latter, we have 
 
 —h+(pi-p)(® -2)-pl=0, -k+(u-9@ -*)-ul=0, 
 both of which the co-ordinate x’ must satisfy. Now let the 
 points be consecutive, and let the second point be situated on 
 a line of curvature the projection of whose tangent on the 
 plane of wy makes an 20 with the axis of w, then k = h tan 0 
 ultimately ; hence, expanding p,, q,, and J in the above 
 equations, substituting for k, and then making h = 0, we find 
 the equations 
 
 (r + s tan 0) (x’ — 3) 
 (s + ¢tan @) (x — z) = tan0 + pq + ¢’ tan 0, 
 
 which must be satisfied by the same values of x’ and tan@ ; 
 hence, eliminating tan @, we must have 
 
 1+ p’ + pq tan 0, 
 
 Le Pe ee ae ee ne 
 “@aas—pqg 9 14+ ¢- (= a)t? 
 or (rt — 8°) (x! — x)? - $1 + p*)t- 2pqs+( + qr} (e — 2) 
 +1+p?4+q°=0,.....(1),; 
 
 the equation which determines z’— x, the projection of a 
 
 principal radius of curvature on the axis of z, so that 
 p=-(® -s)V1l +p ?+q. 
 
 Hence if we find the two values of x’ — x from the above 
 
 equation, and substitute them in the value of p, we shall 
 
205 
 
 obtain the values of the two principal radii of the surface at 
 the proposed point; also knowing z’ — z, the other co-ordinates 
 w and y of the centers of curvature are known from the 
 equations to the normal 
 
 we —xv+p(s —x)=0, y —y+ q(x —2) =0. 
 Cor. The roots of equation (1) have the same or different 
 signs according as r¢—s° is positive or negative; therefore 
 
 the curvatures are in the same or different directions (as we 
 have before seen) according as rf — s’ is positive or negative. 
 
 Also in order that the two radii of curvature may be equal 
 but of different signs, we must have 
 
 (1+ p*?)t-2pqs+ 1+q)r=0. 
 234. In order that the two radii of curvature may be 
 
 equal but of the same signs, we must have (solving equation 
 (1) and putting the part under the radical.equal to zero) 
 
 ,(1+ p*)t-2pqs+(1+q")r}?-4(1 4+ p?+q’) (rt — s*)=0...... (1) 
 or {(l+p)t-(1+q)r+2(1 + 9@°)r -2pqs}? 
 —43(1 +p) (1+ 9°) -— p’g’t (rt —8*) = 0; 
 
 and expanding the first term and reducing, we find 
 S(i+p*)t-(1+q@)r}?4+43 +p’)s—pqr $(1+q°)s—pqt} =0...(2). 
 Let (1+ p*)s —par=pqu, (1+ 9°)s— pat= pau; 
 o Sl +p*)v— (1+ q)uh?+4uv = 0; 
 
 Qu )* 
 b l+p)v-(l+¢ 
 Vice eae tes 
 
 ‘ 4(1+4q°) , 4u 
 =jJ(l+p)v-(C+q)ul? + 4uv— — —~ u? + ——__; 
 
 ee a eae 
 a {tp )o- (144 _ —) a + Canty 4u°=0...(3), 
 
 an equation which can only be satisfied by putting 
 u= OQ, v= 0, 
 
 or (1 + p')s— pgr = 0, (1+q')s — pqt =0, 
 
206 
 
 the conditions that the two radii of curvature may be equal 
 and of the same sign. 
 
 From this process it appears generally that the roots of 
 the equations (Arts. 231, 233) which give the principal radii, 
 and the directions of the principal sections, at any point of a 
 surface are always real; for these roots involve the square 
 roots of the polynomials (1) and (2), which are here shewn to 
 be equivalent to one another, and to (3) which is a form that 
 never can become negative. 
 
 235. The points of a surface at which the two radii of 
 curvature are the same both in magnitude and sign, are called, 
 as has been stated, umbilici. The co-ordinates of such points 
 must satisfy the double equation 
 
 l+p? pq 14+¢ 
 
 5 
 ry s t 
 
 at which we may arrive immediately by observing that, since 
 at such points all normal sections have the same curvature, 
 the expression for the radius of curvature of any normal 
 section, 
 
 V1+pit+g fi +p +2pqm + (1 + g’)m"? 
 y+ 2sm + tm 
 
 R= 
 
 must be independent of the quantity m which particularizes 
 the normal sections; and for this it is necessary and sufficient 
 that the coefficients of like powers of m in the numerator and 
 denominator should be equal ; 
 
 Pep) pgs A 
 
 Also at these points the equation which determines the 
 directions of the principal sections at any point, since it may 
 be put under the form 
 
 1 Yo —- (1 K 
 m + {utero m—-uU 
 PY 
 
 becomes identical, and therefore the directions of the principal 
 sections indeterminate; in fact, all the normal sections having 
 
 = 0, 
 
207 
 
 the same curvature, each of them may be considered as a 
 principal section. 
 
 And the shortest distance between the normal at the um- 
 bilicus and the normal at a consecutive point taken in any 
 direction, will vanish as far as it depends upon small terms 
 of the first order; and for certain directions it will vanish as 
 far as it depends upon small terms of the second or higher 
 orders likewise. For the first member of the equation 
 
 (Art. 231) 
 (h+pl)(1-9)-4&+a90(p-p)=9 (1) 
 
 is the numerator of the shortest distance between the normals 
 at wyx and any adjacent point ~+h, y+k, x47; and ex- 
 panding p,, q,, / in powers of # and k as far as the second, 
 we get 
 
 P=p+dp+tdp, n=qt+dq+ hdq, l=dz+hd's 
 where dp=rh+sk, @p=uh’ + 2vhk + wk’, 
 
 and similarly for the others, so that the first member of (1) 
 becomes 
 
 sh+(p+dp+ 3 d’p) (dx + 4d’z)} (dq+4d’q) 
 —jk+(q+dq+4d'q) (dz + £d’z)t (dp + 4d’p). 
 But if wyz be an umbilicus, all the terms of less than three 
 
 dimensions in h and k are identically zero; therefore writing 
 down only the terms of three dimensions we get 
 
 (h + pdx) d’q + pdzdq —(k + qdz)d'p — qdizdp, 
 which put equal to zero gives a cubic for determining 
 tand=k—h; 
 
 so that in general there will be either one or three directions 
 in which we may pass from an umbilicus to an adjacent point, 
 so that the normals shall have a coincidence of a higher order 
 than for all other adjacent points. If the coefficients of the 
 cubic should be identically zero, we must have recourse to the 
 terms of a higher order in the expansion of (1), and we shall 
 generally determine a finite number of directions relative to 
 which the normals coincide more closely with the normal at 
 the umbilicus, than do the normals at other adjacent points ; 
 
208 
 
 and those directions more especially deserve the name of lines 
 of curvature through the umbilicus. 
 
 236. To find whether a given surface admits of umbilici, 
 we must substitute the values of p, g, 7, 8, ¢, derived from its 
 equation in the double equation 
 
 = i 
 
 and examine whether these together with the equation to the 
 surface can be satisfied by real values of a, y, x. Consequently 
 the number of umbilici of a given surface will in general be 
 limited ; only if the double equation should be reducible to a 
 single distinct equation, then that equation joined to the 
 equation to the surface will determine a curve on the surface 
 every point of which will be an umbilicus, and which is called 
 a line of spherical curvature, because about each of its points 
 the surface is uniformly curved like a sphere. 
 
 237. If through all the points of a line of curvature 
 KPP (fig. 68) we draw normals to the surface, since every 
 two consecutive oues intersect, they will form a developable 
 surface whose edge of regression will be the locus of the 
 centers of the first curvature relative to KPP’. Proceeding 
 in a similar manner for each line of the same curvature LQQ’, 
 &e., we shall obtain a series of developable surfaces of which 
 the edges of regression will form by their assemblage a surface 
 which is the locus of all the centers of the first curvature, 
 and to which all the normals will be tangents. Also this 
 surface will have a second sheet which will be the locus of the 
 centers of the second curvature, resulting from the edges of 
 regression of the surfaces generated by the normals along the 
 lines of the second curvature RPQ, MP’Q’, &c.; and which 
 will be touched by the same normals as the other. 
 
 To obtain the equation to this surface, we must eliminate 
 w, y, = between the equation which gives x’ — x, the projection 
 of a principal radius upon the axis of x (Art. 233), and the 
 equations to the normal. When the two sheets intersect, 
 their intersection will be the locus of the centers relative to the 
 line of spherical curvature of the surface under consideration. 
 
APPENDIX. 
 
 Tur following Problems, distributed into Sections cor- 
 responding to those into which the Treatise is divided, will 
 furnish occasion for applying the results in the Text as they 
 are successively obtained. 
 
 ProspiteEMs on SeEctTion I. 
 
 1. To find the equation to a plane considered as generated 
 by a straight line always passing through a fixed point and a 
 fixed straight line. 
 
 Let , y, x be co-ordinates of P any point in the generating 
 line CQ (fig. 7) passing through the fixed point C in the axis 
 of zx, and the fixed straight line AB in the plane of wy. Let 
 x’, y’ be co-ordinates of Q, and a, b, c the lines OA, OB, OC; 
 then 
 
 oe yf ev ON -c—x# *y- ON. ¢-—z 
 
 =+>1;, bout.—=—— = De es ee 
 
 Fh A, a OR Cc y OQ G 
 ~+0=1--, the equation required. 
 
 2. The equation to the plane generated by a straight 
 line that always meets a line whose equations are v7 = mz +a, 
 y=nx-+b, and that remains parallel to a line through the 
 origin whose equations are w = m’z, y =n’, is 
 
 e-mx—-a y—n'zs—b 
 
 m — m' n—n' 
 
 3. To find geometrically the distance of a point from a 
 plane. 
 E4%, 
 
210 
 
 Let a’, y’, x’ be the co-ordinates of the given point P 
 
 (fig.9), s=Av+ By+c the equation to the given plane ABC. 
 
 Through P draw a plane GAH perpendicular to the trace 
 AR; then this plane is perpendicular both to the given plane 
 ABC, and to AOB, and contains the ordinate PN. Let GH 
 be the intersection of the two planes meeting PN in Q; draw 
 PR perpendicular to GH. Then PR is perpendicular to the 
 plane ABC; and PR = PQsin PQR = PQcos GHK. 
 
 But PQ=PN- QN =2'- (Aa'+ By' +0), 
 
 since Q is a point in the plane ABC, for which w=a’, y=y ; 
 1 
 and cosGHK = €Art. 29); 
 J/i+ 4+ B 
 
 e— Aa'— By -c_ 
 2/14 A+ Be 
 the radical being taken with that sign which makes the whole 
 expression positive. 
 
 .. PR, the required distance, = 
 
 4. To find the equations to a straight line which cuts 
 perpendicularly each of two straight lines not in the same 
 plane, whose equations are given. 
 
 Let 
 
 / , 
 L=MN2FZ+Aa VC=mese+a : 
 \, : | be the equations to the 
 y=nst+b y=ns+b 
 given lines; then if x= da + By+e be the equation to a 
 plane parallel to them both, dm+Bn=1, Am'+ Bn'=1, 
 
 (Art. 23) 
 
 , , 
 n—-n m— mm 
 = Saree O B=————_, 7° 
 mi —- mn Mm — M7 
 
 Let x = A’a + By +c’ be the equation to a plane which 
 contains the former of the given line and also the required 
 line, and which therefore (Art. 41) cuts the plane parallel to 
 both the given lines at right angles; therefore (Art. 23, and 35) 
 
 AA'+ BB +1=0, Am+Bn-1=0, A’'a+Bb+c=0; 
 
 hence c= — d'a— B’b, 
 
211 
 
 and A'(dn - Bm)=-(B+n), B'(Bm-An) = -(4 4m); 
 “. =A’ (w@—a) +B (y - bd), 
 or = (dn - Bm) = - (e@- a) (B+) + (y—-b) (44m), 
 or, substituting the values of 4 and B, 
 x Ym (m'— m) +n (n'— n)} = (w — a) §m'—-m+n(m'n - mn')} 
 + (y —b) j{n’-n+m(mn'—m'n)}, 
 is the equation to the plane containing the former of the given 
 
 lines and the required line. 
 
 Similarly, the equation to the plane containing the latter 
 of the given lines and the required line, by interchanging 
 m,n, a, b, and m’, n’, a’, b,, is 
 
 = }m'(m —m') +n (n-n')h =(w-a’) {m-m'+n'(mn'- m'n)} 
 
 + (y-0)}n-n'+m'(m'n—-mn')t. 
 These two equations determine the position of the line re- 
 quired, since they are the equations to two planes each of 
 
 which contains it; the equations to its projections may be 
 
 found by Art. 20. 
 
 5. To find the length of the shortest distance between 
 two straight lines whose equations are given. 
 
 If a plane be drawn through each of the lines parallel 
 to the other, the perpendicular distance of these planes which 
 will of course be parallel to one another, will be the shortest 
 distance of the lines (Art. 41). 
 
 8 
 I| 
 
 Let 
 
 meta) wvemsz+a’ 
 ) : py be the equations to the 
 y= no +b y=nszt+b 
 
 lines, x = da + By +e the equation to the plane which is 
 parallel to the second and contains the first, 
 
 - Am'+ Bn'=1, Am+Bn=1, da+Bb+c=0, 
 
 which give for A, B, c, the values 
 n'—n m’ — m 
 A =————— , B= - ——,, c=- Aa-— Bb. 
 
 , a > , , b) 
 mn—-m nr mn — mn? 
 14—2 
 
212 
 
 Similarly, the equation to the plane which is parallel to the 
 first and contains the second, is = dva+ By+c’, where 
 A, B have the above values, and c’ = — Aa’ — Bb’; and the 
 lengths of the perpendiculars dropped upon these planes from 
 the origin are respectively (Art. 37. Cor.) 
 
 — Cc —C 
 
 V1 + ANE an lee 
 
 — (c'-c) A (a’— a) + B(b'-b) 
 nat at aera /1 + A+ B 
 therefore, substituting for 4 and B their values, the shortest 
 distance between the two lines 
 
 (n’— n) (a’— a) — (m’'—™m) (b’— 5) 
 V (m'— m) + (n'=n)? + (mn'— m'ny? ; 
 
 Cor. When c in the equation to the first plane, and 
 
 c the corresponding quantity for the second plane, have 
 different signs, the origin is situated between the two planes ; 
 and therefore the distance of the planes will be found by 
 
 taking the sum of the perpendiculars. When the shortest 
 distance vanishes, 
 
 (n’— n) (a’— a) = (m’- m) (b'— b), 
 which is the condition in order that the lines may intersect, 
 found in Art. 22. 
 
 whose difference = 
 
 6. Having given the lengths, the least distance, and the 
 inclination of two opposite edges of a tetrahedron, to find its 
 volume. 
 
 Let 4 (fig. 11) be the vertex, and BOC the base of any 
 tetrahedron; complete the parallelogram Cy, and join Ay; 
 let Nw cut each of the opposite edges 4B, OC, at right 
 angles; then OC is parallel, and Nw is perpendicular to the 
 plane ABy. Hence, since they are on the same base, and 
 their vertices are in a line parallel to the plane of the base, 
 
 tetrah. dy BO = tetrah. dyBu = 4 AB. By.sin ABy. 4 Ne; 
 but tetrahedron Ay BO = tretrahedron AOBC, 
 
213 
 
 since they have a common vertex, and equal bases in the 
 same plane ; 
 
 ‘, tetrahedron AOBC = 1 4B. OC.Nvw.sin 8, 
 where @ denotes the angle between AB and OC. 
 
 7. A sphere may be described touching each of the six 
 edges of a tetrahedron, provided the sum of every two 
 opposite edges be the same. 
 
 8. A cube may be cut by a plane so that the section 
 shall be a square whose area is to that of the face of the cube 
 as 9 to 8. 
 
 9. Ifa plane be drawn so that the sum of the perpen- 
 diculars let fall upon it from m given points a’y’s’, vy’, &c. 
 shall be always equal to a given line A, then it will always 
 
 touch a sphere the co-ordinates of whose center, and whose 
 1 
 
 radius, are respectively F th of the quantities =(#’), = (y’), 
 
 >= (z’), and R. 
 
 10. The shortest distance between a diagonal of a cube 
 and an edge which it does not meet is a+»/2, @ being an 
 
 edge. 
 
 11. If @ be the inclination of two planes /a+my+nz=a, 
 lv + m’y + nx =, the distance of their line of intersection 
 
 wns 1 > ik ea = Aes He 
 from the origin = arp wo 2a(3 cos 0 + 2°. 
 
 ProspLEMs ON Section II. 
 
 The following Problems furnish examples of finding the 
 equations to surfaces from the given geometrical mode of 
 describing them. As the simplicity of the result will in 
 every case greatly depend upon properly selecting the posi- 
 tions of the co-ordinate axes, some instances are here intro- 
 duced with that special object. 
 
214 
 
 1. If a straight line have always three given points in 
 three fixed planes at right angles to one another, to find the 
 locus of any other point in it. 
 
 Let the planes be taken for the co-ordinate planes, and 
 let the line meet them in the points 4, B, C, (fig. 23); 
 and let P be the describing point the co-ordinates of which 
 are On=a, NQ=y, PQ=z2. Alsolet PA=a, PB=b, 
 PC=c, ZACa=i, £Cbx=0, Ca being the projection of 
 CA on wy. Then 
 
 z=csini, w=acosicos#, y=bcosisin®@; 
 
 2 
 
 ne y Pe i aie 
 —+—+— = 005 @+sin*2z=15 
 C 
 
 a* b* 
 
 hence the locus of P is an ellipsoid, 
 
 2. To find the locus of the intersection of two planes 
 drawn through two given lines, so as always to be perpendi- 
 cular to one another. 
 
 Let 44’, BB’, (fig. 24) be the two given lines, AB 
 their shortest distance which make the axis of x; also take 
 the plane bisecting AB at right angles for that of «y, and 
 the line bisecting the angle between the projections Oa, Ob 
 of the lines, for the axis of « Then if O04 = OB=ce, and 
 tanaOw=m, the equations to the lines 4d’, BB’, are 
 Y=oMe, ®=C}; y= —Me, £=—C. 
 
 Let zs = Av + By + € be the equation to the plane passing 
 
 through 44’; .. d+ Bm=0; 
 w-en A (v2) Bi PAS BAe ‘hit Se 
 m me—y me—-—y 
 
 Similarly, if * = A’w + B’y -c be the equation to the plane 
 passing through BB’, d’- B'm=0; 
 
 A eae: ppp TOS. 
 
 me+y Lina ty) 
 
215 
 But, because the planes are at right angles to one another, 
 
 AA + BB+1=0, ©: ples aaa SES ee Tt eI), 
 (mv) -—y (mx)? —y 
 
 or (m* —1)2°+4+ ma’ — y? = (m? — 1) ce’, 
 
 which, since only one of the coefficients is negative whether 
 m> or <1, represents a hyperboloid of one sheet. 
 
 Cor. It may be easily shewn that each of the lines 
 AA’, BB’, lies in the surface, and that a plane perpendicular 
 to either of them will cut it in a circle. If the lines are 
 parallel, m=0, and vy’ + 3’ =c*, which represents a cylinder 
 as it ought. 
 
 3. To find the locus of a point which is equidistant 
 from two lines given in space. 
 
 Retaining the same axes and notation as in Prob. 2, it 
 will be found that the surface is a hyperbolic paraboloid 
 having for equation 
 
 may + (14+ m’)cz=0. 
 
 4. Two points move in straight lines with uniform 
 velocities, to find the equation to the surface generated by 
 the straight line which joins them. 
 
 Retaining the same axes and notation as in Prob, 2, 
 and besides denoting by a, a’, the initial values of w for 
 the points moving in 44’, BB’, respectively, and by 7 : 1 
 the ratio of their velocities, it will be found that the equa- 
 tion to the surface is 
 
 (1 +1)c(cy—mazx)—(n—1)c(mea—zy)=m(a-na)(c? —32"). 
 5. ‘To find the equation to the surface generated by a 
 
 straight line which constantly passes through three fixed 
 straight lines. 
 
 Let a parallelopiped be constructed, having its three 
 edges (of which no two either intersect or are parallel to 
 
216 
 
 one another) in the three given lines; take its center for 
 origin and lines parallel to its edges for the axes; then the 
 equations to the given lines will be 
 
 Let the equations to the generating line in any posi- 
 tion be 
 
 a—h —k 
 v=mzt+h, y=nzr+k, a“ atl ; 
 m 
 
 then since it meets each of the given lines, we have three 
 equations of condition, viz. 
 a—h b+k 
 
 —-a@=mc+h, b=-ner+hk, = ; 
 m n 
 
 and it remains to eliminate m, n, h, k, between these and 
 the equations to the generating line. The result is 
 
 ayz+bzx+cxy+abe=0. 
 
 Cor. If the signs of the quantities a, b, c be changed 
 this equation is not altered; therefore the same surface would 
 be generated by a straight line constantly passing through 
 three different straight lines whose equations are 
 
 v= ‘t hs Mat 
 3 3 * 
 s=-€ Zi ,uc y= 6 
 The surface, as we know, is a hyperboloid of one sheet; the 
 latter therefore is the only surface that can be generated in 
 the manner described above; the origin is its center, and the 
 
 axes of the co-ordinates are situated in the surface of the 
 conical asymptote. 
 
 6. To find the surface generated by a line which always 
 intersects two given lines, and is parallel to a fixed plane. 
 
 Let the fixed plane be taken for that of yx, and let 4B 
 (fig. 24), the line joining the traces of the given lines on that 
 
217 
 
 plane, be taken for the axis of x, and its middle point for the 
 origin. Also, let the plane of wy be parallel to the two given 
 lines, and the axis of w bisect the angle between their pro- 
 jections made parallel to Oz. Then referred to this system of 
 oblique axes the equations to the given lines are 
 
 =Me#, Z=C3 Y= —-MX, F=—=—C. 
 > 
 
 Let v=a, y=nz+b, be the equations to the generating 
 line since it is parallel to yz; then since it intersects each 
 of the given lines 
 
 ma=ne+b, —-ma=—nc+b; 
 
 “. b=0, and ma=ne, or eliminating a and n, mzx=cy 
 is the required equation, representing, as we know, a hyper- 
 
 bolic paraboloid. 
 
 7. To find the surface generated by a straight line con- 
 stantly meeting three given lines that are parallel to the same 
 plane. 
 
 Take one of the given lines for the axis of wv, and a line 
 intersecting it and parallel to another of the given lines for 
 the axis of y, so that the plane of wy is that to which the 
 three given straight lines are parallel; and for the axis of x 
 take a line joining the origin with the point in which the third 
 given line meets the plane containing the axis of y and the 
 line to which it is parallel. Then the equations to the three 
 
 given lines are 
 y = 0 v= 0 Y= Ma 
 ane set =k | 
 
 and the equation to the surface will be found to be 
 
 kyz+m(h—-k)ex=hky, 
 
 representing a hyperbolic paraboloid since the surface has not 
 a center, and is generated by a straight line that does not 
 continue parallel to itself. No two of the fixed lines must 
 be in the same plane; for if so, two planes will be generated, 
 viz. that containing the two lines and meeting the third 
 at an infinite distance, and that passing through the point of 
 intersection of the two lines and the third line. 
 
218 
 
 8. The locus of a point, whose distance from a fixed 
 point is always equal to times its distance from a fixed line, 
 
 ; y ne NC 
 is a spheroid with semi-axes ——_,, —=—., c being the 
 1-n V/1 — n° 
 
 perpendicular distance of the fixed point from the fixed line. 
 
 Let A (fig. 6) be the given point, HA-=c the perpen- 
 dicular from it on the given line 47’; produce 4H to O so 
 that 40 = ——., and consequently HO = — ;- Draw Ox 
 
 —n 
 parallel to 47’ for axis of x, and take OA for axis of wv, and 
 let P with co-ordinates wv, y, x be a point in the locus so that 
 
 Ei ee), aE 
 
 ne 2 2 
 ( - 2) +y +n’, MA =an’ ( _- 2) +n'y’; 
 lL —-n 
 : be Set ee 
 or CU itp) Sia ste Stag (I sartt) rag oe 
 —n 
 
 which represents a spheroid with the semi-axes above stated. 
 
 9. To find the locus of a point whose distance from a 
 given point always equals m times its distance (measured 
 parallel to a fixed plane) from a given line. 
 
 Take a plane through the given point parallel to the 
 fixed plane for the plane of wy, and a plane through the given 
 line perpendicular to the fixed plane for that of vg, and sup- 
 pose the axis of y to contain the given point, then 
 
 a+ (y—b)? +2? = n'y? +n? (mz+a- 2)? 
 
 is the required equation, v= mz +a being the equation to 
 the given line in the plane of vz. 
 
 10. The locus of a point whose distance from a fixed 
 plane (that of wy) always equals its distance from a line 
 inclined at an angle a to that plane, has for its equation 
 
 y? — vy sin2a + (a — 2°) sin? a = 0. 
 
219 
 
 11. To find the equation to the surface of a cable-ring, 
 having given a and ¢ the radii of its inner and outer boundaries. 
 Also to shew that the section made by a plane touching the 
 inner boundary is a Lemniscata when ¢ = 3a. 
 
 Let ACB a diameter of a circle whose center is C meet 
 Oz, a line in the plane of the circle, at right angles in O; 
 to find the equation to the surface generated by the revolution 
 of the circle about Oz. Take ON, NP co-ordinates of P a 
 point in the circle, then 
 
 PN’ = AN.NB=(OA-ON).(ON — OB), 
 or 3° = (w—a)(c - @) 
 
 is the equation to the generating curve in the plane of za; 
 therefore (Art. 52) the equation to the surface generated is 
 
 = (/ a + y —a)(e- J v? + y). 
 
 To determine the section made by a plane touching the 
 inner boundary let v = a, then 
 
 v= (fat y—a)(c- Vaty’), 
 or (2° + 9’)? = (c + a) $(c — a) y’ — 242°}, 
 which becomes the equation to the common Lemniscata if 
 
 Cras SG: 
 
 12. If the axes of a hyperboloid become evanescent, 
 always preserving a constant ratio to one another, the surface 
 is changed into its asymptotic cone, 
 
 Let the equation to the hyperboloid be 
 
 2 2 2 
 
 a, b, ec being certain given values of the semi-axes; and sup- 
 pose the semi-axes to diminish continually, but always to be 
 proportional to a, b, c; then they may be represented by na, 
 nb, me, where m is a common multiplier continually growing 
 
220 
 
 less; therefore the equation to the surface in its successive 
 changes will be 
 
 a y° x? x 
 oe ie ie OTe 
 na nb ae CD 
 
 3 a 
 Cc 
 
 therefore, making 2 = 0, or the semi-axes na, nb, ne, evan- 
 escent, the equation to the surface becomes the same as that 
 to its conical asymptote, viz. 
 
 ei it 
 
 13. If two fixed lines pass through a point, the locus of a 
 third line passing through the same point and making angles 
 a, a’ with the former, such that tanda.tan4q’ is invariable, 
 is a cone of the fourth order. 
 
 14. To find the locus of a circle whose center is the 
 origin, and which always passes through the axis of z, and 
 through an ellipse whose equation is a’y’ + b’a* = a®b’. 
 
 15. If three lines mutually at right angles and passing 
 through a fixed point C intersect the surface of a sphere, 
 then the center of gravity both of the triangle formed by 
 joining any three points of intersection, and of the pyramid 
 having C for its vertex and that triangle for base, will lie in 
 spherical surfaces described about C as center with radii 
 
 vas ae 2 ay 22 _ oO? 
 =3V 3r 2d and 44/8r 2d 
 
 respectively, » being the radius of the sphere and d the 
 distance of its center from C. 
 
 16. If the paraboloid Ja + Uy? =Jlz be cut by planes 
 through its axis, the directrices of the sections will lie in a 
 surface whose equation is 4% (Ja + I'y”) + Ul’ (a + y’) =0. 
 
 17. If one of the co-ordinates of an ellipsoid be produced 
 so that the part produced may equal the sum of the other two, 
 its extremity will trace out an ellipsoid of the same volume as 
 the original one. = 
 
221 
 
 18. The locus of a circle that always touches the axis 
 of x at the origin, and also passes through a fixed straight 
 line ay + bw = ab in the plane of wy, has for its equation 
 
 x (ay tbe) +(e + y’) (ay + bw - ab) = 0. 
 
 Prospitems on Section III. 
 
 We shall here give some applications of the principal 
 results obtained in this section, relative to the projections 
 of plane surfaces, the transformation of co-ordinates, and the 
 intersections of surfaces by planes. 
 
 1, The pyramid whose vertex is the origin, and base any 
 plane surface, is equal to the three pyramids whose common 
 vertex is any point in the plane of the surface, and bases its 
 three projections on the co-ordinate planes. 
 
 Let d, a perpendicular from the origin upon the plane in 
 which the surface lies, make angles a, 3, y with the axes of 
 X,Y, %; also let A denote the area of the surface, and a, y, z 
 the co-ordinates of any point in its plane, 
 
  wcosat+ycosf + cosy =d. 
 Hence, multiplying by 4.4, 
 
 Acosa.tx+AcosB.4y+ Acosy.4%=A 
 
 ve 
 
 or, dy. + Ay. + Ay. 
 
 Go | ee 
 1} 
 ae 
 
 which (since the volume of any pyramid is equal to the 
 area of its base multiplied by one third of its height) ex- 
 presses the property announced. The quantities d,, 4,, 4, 
 will be positive or negative according as a, 3, y are acute 
 or obtuse. 
 
 2. To find the sum of the projections of any number of 
 areas upon a given plane. 
 
222 
 
 Let D’, D", &c. denote the areas; a’, By’; a’, By’; 
 &e. the angles which the perpendiculars upon their planes 
 from the origin make with the rectangular axes of a, y, % 
 produced in the positive directions. Also, let a, B, y be 
 the angles which a perpendicular to the given plane makes 
 with the same axes. ‘Then the areas must be multiplied 
 respectively by the expressions 
 
 f , , 
 cos a cos a’ + cos (3 cos [3 + cos vy cos +y 
 cos a cos «+ cos ( cos 3” + cosy cosy”, &e., 
 
 which are the values of the cosines of the angles between the 
 plane of projection, and the planes in which the areas lie. 
 The sum of these products will be 
 
 S' = cos a (D' cos a’ + D" cos a” + &e.) 
 + cos 3 (D' cos 3’ + D” cos 3" + &c.) 
 + cosy (D' cosy’ + D" cosy" + &e.) ; 
 
 or, if we denote the sums of the projections on the co-ordinate 
 planes of yz, za, vy by S,, S',, S, respectively, 
 
 S' = S, cosa + S,, cos B + S, cos vy; 
 
 which enables us, from knowing the projections of any areas 
 upon three planes at right angles to one another, to find 
 the sum of their projections upon any plane whose position 
 is given relative to the same planes. In forming the values 
 of S,, S,, S, regard must of course be had to the signs 
 of cosa, cosa’, &c. 
 
 3. To find the position of the plane on which the sum 
 of the projections of any number of areas is a maximum. 
 
 Let S,, S,, S, denote the sums of the projections of 
 the areas on the co-ordinate planes of yx z#, wy respec- 
 tively; S,, S',, S, the sums of the projections on three 
 other planes y'2’, s'v’, a'y’ at right angles to one another 
 having the same origin, 
 
223 
 
 . Sy = S, cos wae + S, cos v’y + S, cos v’x, 
 t ’ , , 
 S, = S,cosyxv+S,cosyy + S, cosy x, 
 
 S,, = S, cos z'v + S', cosx’y + S, cos 2%. 
 
 Hence, squaring and adding and taking account of the equa- 
 tions of condition to which the cosines are subject, 
 
 y2 2 2 2 2 2 
 Sy + 8, + S, = S, +S, + 8;. 
 
 Hence, if the planes of y's’, 2’a’, ay’, be so situated that 
 the sums of the projections on two of them vanish, the sum 
 of the projections on the third will be the greatest possible ; 
 let this be the plane of y’z’, then S,=0, S,=0, and 
 
 the greatest sum S'= Sy = V/s? + 5) + S%. 
 
 But if we had begun by supposing the projections on 
 the planes of y's’, 2a’, ay’ given, we should have had 
 the equation 
 
 S, = Si cos va’ + S,,cos xy’ + S, cos wz’. 
 In this case therefore, S$, = S, cos va’; similarl 
 >) @ oH >) 3 
 S, = S,cosya’, S,= Sy cosza'; 
 
 which three equations determine the position of the plane (or 
 rather of its normal, viz. the axis of 2’) on which the sum 
 
 of the projections is a maximum, since S‘y = JS? as SY + 82. 
 
 Cor. Let S” denote the sum of the projections on a 
 plane whose normal makes angles a’, 6’, y’ with the axes 
 of w, y, x, and an Z=8 with the normal to the plane of 
 greatest projection, and let the latter normal make angles 
 a, 2, y with the axes; 
 
 .. cos @ = cos a cos a’ + cos 3 cos 3’ + cosy cosy’ 
 = S', cosa’ + S,, cos 3’ + S', cos ry’ ~~ S 
 S SS” 
 .*. S" = S'cos 0. 
 
224 
 
 Hence, the sum of the projections vanishes on all planes per- 
 pendicular to that on which the sum is a maximum. 
 
 4. To find the area of the surface of any portion of a 
 right cone on a circular base. 
 
 Let ACM (fig. 33) be a section through the axis of the 
 cone perpendicular to the cutting plane CPB; 
 
 AC=c, AB=c, LCAB=B; 
 
 also let Cpb be the projection of CPB on a plane perpen- 
 dicular to the axis, then 
 
 Cb = (c +c’) sind B = major axis of projection ; 
 
 and 4/ce' sin 1 =4 minor axis both of CPB and Cpb, since 
 it Is a mean ioe between the perpendiculars dropped 
 from B and C upon the axis of the cone; 
 
 es (c+c) Vee 
 
 . the area of projection Cpb = z sin® : : 
 But Cpb is the projection of the conical BUMAEE CABP, 
 every portion of which is inclined at an Z = 90°- +B to the 
 plane CM, 
 B(ce+ (c-+¢) See / ce 
 
 ‘, area of surface CABP = r sine TC Ba 
 
 5. Three straight lines mutually at right angles and 
 meeting in a point, constantly pass through a plane curve 
 of the second order; to find the locus of their point of 
 intersection. 
 
 Let Aa’?+ BY =C 
 
 = 0 
 
 h, k, & the co-ordinates of the point of intersection of the 
 three lines; and let the curve be referred to that point as 
 origin, and to the three straight lines as axes; let the co- 
 sines of the angles formed by the new axes of a’, y’, 2’, with 
 the axis of w, be denoted by m, m, 7; and similarly for the 
 
 be equations to the plane curve, 
 
225 
 
 axes of y and zs; therefore by the formule of Art. 90, the 
 transformed equations to the curve are 
 
 A(ma'+ ny + rx’ +h)? + B(m'a' + n'y + 7'2'4+ kP=C, 
 mae + n'y +r es +1=0. 
 In order that the axis of a may meet the curve, these 
 
 two equations must agree in giving the same value for 2’, 
 when y’ and x’ =0; 
 “ y 
 “. A(ma’ +h)?+ B(m'e' + ky =C, ma’ +1= 0, or a’ = —-—; 
 m 
 
 hence, substituting this value of a in the former equation, 
 and proceeding in a similar manner with respect to the axes 
 of y’ and 2’, we have 
 
 A(m”h— ml)? + B(m’k —- m'l)? = Cm’”, 
 A (n"h — nl)? + B (nk - n'l)? = Cn", 
 A (rh — rl)? + B(r"k - 7'l? = Cr”. 
 
 Therefore, adding these equations together and taking account 
 of the equations of condition to which the quantities m,n, &c. 
 are subject, 
 
 A(h?+l?)+B(K+P)=C, 
 the equation to the locus; which is therefore a surface of 
 the second order concentric with the curve. 
 Cor. If the given curve be an ellipse, this equation 
 becomes, replacing h, k, 1 by «#, y, 2, 
 Ba” + a’y’ + (a? + b*)2? = a’? b’," 
 that to an ellipsoid; if a hyperbola, the equation to the sur- 
 
 face is 
 ba? — a’y? — (a — b*) 2? = a’d’, 
 
 that to a hyperboloid of one or two sheets, according as 
 the transverse axis of the hyperbola <or > the conjugate axis. 
 
 In the case of the ellipse, if we remove the origin to 
 15 
 
226 
 
 the extremity of the major axis, by writing w — a for a, the 
 equations become 
 
 we | 02a : ‘ : 
 
 ——-— + = = 0, Ba +a’y’ + (a? +0°)2 = 2ab'a; 
 
 a a b- 
 now let b°>=2ap —p’, p being the distance of the focus of 
 the ellipse from its vertex, and make a infinite ; 
 
 ~YyY=4pa, y+" =4pH; 
 
 which shew that when the curve is a parabola, the locus 
 required is the paraboloid which it would generate by re- 
 volving about its axis. 
 
 6. Three straight lines mutually at right angles and 
 meeting in a point are applied to a surface of the second 
 order; to find the locus of their point of intersection. 
 
 Let the equation to the surface be da’ + By? + Cx? =D; 
 then referring it to the three straight lines as axes and 
 to their point of intersection as origin, the transformed 
 equation is 
 
 A(ma’+ ny +rz' +h) + B(m'a! + n'y +7°x' +k) 
 +C(m'e' + n'y +7's' +l)’ =D. 
 Now making successively y’, 2’, =0, a’, 2’, =0, a’, y', =0, 
 in order to determine the points where the axes of a’, y’, 2’ 
 meet the surface, and calling Ah?+ Bh? + Cl’—- D=G, we have 
 
 v?(m?A + m?B + m'?C)+ 220 (mAh +m Bk +m’Cl) + G=0, 
 y? (nA +n? B+ n'?C) + 2y'(nAh+n' Be +n" Cl) + G=0, 
 8°? At rP B+ r?C)4+2ek(7Ah +r Bk + 7° Cl) + G=0. 
 
 But since the axes of a’, y’, x’ touch the surface, the two 
 points in which any axis meets it coincide in one; therefore 
 the roots of each of the above equations are equal, and the 
 first members are perfect squares ; 
 
 “(mA + mB + m'C)G = (mAh + m'Bk + mC), 
 (A+ n?B+n?C)G =(nAh +n'Bk +n’ Cl)’, 
 Ate Bir * C)Ge (Ahk +r Bk +4 Cly: 
 
2947 
 
 Hence, adding together these equations, and taking account 
 of the equations of condition to which the quantities m, n, &c. 
 are subject, 
 
 (44 B+ C)G= Ah? + Bi’ + CP, 
 or, restoring the value of G, and reducing, 
 A(B+C)h?+ B(A+ OC) 4+C(44 BYP =(A4+B+C)D, 
 
 the equation to the locus, which is therefore a surface of the 
 second order concentric with the proposed one. 
 
 Cor. For the ellipsoid, replacing h, k, 1 by a, y, 2, 
 this equation becomes 
 (P+e)a+(P+ec)y + (2 +0) = 0B? + ae? + b’c’, 
 x’ y” 3? 
 
 or Se aetee et 
 & s- 
 
 if g, g’, g” denote the three altitudes of the triangle formed 
 by joining the vertices of the ellipsoid (Cor. Art. 82). 
 
 =)15 
 
 ae) 
 
 If we remove the origin to the extremity of the semi- 
 axis a, and then make a infinite, the equations become 
 2 Pad 
 
 7 4 en =0, Yrs a4(pt pat app’; 
 2p ° 2p 
 
 if therefore the given surface be a paraboloid, the locus 
 required is a paraboloid of revolution. 
 
 If the given surface be a hyperboloid of one sheet, making 
 c’ negative, the equation to the locus is 
 
 (P-—c)e + (@—e)y + (C84) =e -e(a +8); 
 
 the locus is therefore an ellipsoid when c, the imaginary semi- 
 
 ‘nat ie ab , 
 axis, is less than —,———— and of course less than either 
 
 J/ a +b? 
 
 of the real semiaxes; a point, viz. the center, when 
 ab F 
 
 c¢ = ————; imaginary, till c = b the less of the real semi- 
 a? + b° : 
 
 15—2 
 
228 
 
 axes; a hyperboloid of two sheets till ce =@, when it becomes 
 a hyperbolic cylinder; and a hyperboloid of one sheet for 
 all future values of c, i. e. whenever the imaginary axis is 
 greater than either of the real axes. 
 
 If the given surface be a hyperboloid of two sheets, 
 making 6b? and c’ both negative, the equation to the locus is 
 
 (PP +c%)a7 4+ (P-a) + (CP -—a)P’=CWC (C+) - Be’, 
 
 which is imaginary, as long as a the real semiaxis is less 
 
 be : 
 than ir: af ‘and of course less than either 6 or ec; a 
 ge Oe, Cc 
 : be ees 4 
 point, when a = ————-; afterwards an ellipsoid, till a =e 
 
 \/ b+ & 
 the less of the imaginary semiaxes, when it becomes an 
 elliptic cylinder ; a hyperboloid of one sheet, till a = 6 when 
 it becomes a hyperbolic cylinder; and for all future values 
 of a, a hyperboloid of two sheets, i.e. whenever the real 
 axis exceeds each of the imaginary axes. 
 
 e a e e e e ] 
 Also, if in the original equation we put D=0, 4 =—, 
 = 
 
 I 1 ; : 
 B= Be? C= ai we have the equation to a conical sur- 
 face of the second order, and the equation to the locus be- 
 comes 
 (0? + c*) a’ + (a? + c*)y? + (a’ + b’)2* = 0, 
 which is also the equation to a conical surface of the second 
 order; one, or two of the quantities a*, b’, c® being negative. 
 
 7. To find the equation to the section of an oblique 
 cone on a circular base. 
 
 Let the diameter of the base on which a perpendicular 
 from the vertex falls, be taken for the axis of « Then 
 (Art. 49) making 6 = 0, the equation to the surface is, putting 
 a-2r=a, 
 
 cy” = (ex — az) (er+ a's - ex); 
 
229 
 
 therefore, the equation to the section made by a plane through 
 the axis of y inclined at an 26 to the plane of the base, is 
 
 cy” = (ecos 8 — asin@) }2cra’ + (a’ sin@ — ecos@)a"?. 
 
 Cc 
 a! 
 
 As long as tan@ lies between - and , or 8 between SAB 
 a 
 
 and SEB (fig. 34), 7. e. as long as the plane cuts both sheets 
 of the cone, the section is a hyperbola. 
 
 When tan @ = an or the plane is parallel to ES‘, the 
 a 
 
 section is a parabola. In other cases it is an ellipse, which 
 becomes a circle when 
 
 c’ = (c cos 8 — asin 8) (c cos 8 — a’ sin @), 
 or, c’?+c’ tan’@ =c’—c(a+a) tan@ + aa tan’@; 
 c(a +a’) 
 
 . tan@=0, or tan@ = —— ie 
 aa —c 
 
 The first value of 0 gives the base of the cone ; the 
 second gives 
 
 tan 9 = tan (SAB + SEB), 
 or if DAy be the plane of the circular section, 
 DAE = SAB + SEB, «. DAS = BES; 
 
 that is, the cutting plane makes the same angle with one 
 side of the principal section of the cone, that the base does 
 with the other. ‘This circular section is called the subcon- 
 trary section. 
 
 8. The sum of the squares of the reciprocals of any 
 three semidiameters of an ellipsoid mutually at right angles, 
 is equal to the sum of the squares of the reciprocals of the 
 semlaxes. 
 
 Let a, B, y be the angles which any semidiameter r 
 makes with the axes, then the co-ordinates of its extremity 
 arev=rcosa, y=rcosP, =7 cosy; 
 
 1 cosa cos" — cos" 
 
 a = ae eae ++ : 
 r a b? Cc ? 
 
230 
 
 and forming similar equations for 7’ and 7’, and adding, since 
 the semidiameters are mutually at right angles, we find 
 1 vk 1 1 1 sity 
 APN Ota Fe 
 
 9. Ifa sphere be placed in a eee of revolution, 
 the section of the paraboloid made by any plane touching the 
 sphere is a conic section having the point of contact for a focus. 
 
 ] 
 eT 
 ph inte 
 
 10. The eccentricity of any section of a paraboloid of 
 revolution is equal to the cosine of the inclination of the 
 cutting plane to the axis, 
 
 11. To shew that among the constants m, n, r &c. 
 employed in passing from one rectangular system of co- 
 ordinates to another, the following relations hold ; 
 
 man — nr, m =n"r—nr"', m’=nr —n'r; 
 and similarly for n,n’, n”; andr, 7’, 7’. 
 
 Eliminating y’ and 2’ from equations (3) Art. 90 we get 
 a’ = (n'r" — n"7') a + (nr —nr")y + (nr -—n'r)z 
 =rAma«+rm'y +rm’sz from (5). 
 
 Aman’ —-n"r, rm =n"r —nr”’, Am" =n —n'7; 
 N= (nr — n"7')? + (nr — nr’)? + (n7! — n'ry? 
 = (n? 47? +n’) (r+ 7? 4 7) sin? 6 (Cor. Art. 30) 
 
 = 1,0 being the angle between the axes of y’ and 3’; and 
 “, sin @=1. HenceA= = 1, and the proposed results are esta- 
 blished. 
 
 12. ‘To shew that 
 (mm'm’)? + (nn'n"”)? + (rer)? = (mar)? + (m'n’s’)? + (Mm nr’)? 
 we have (mm’m”)? 
 = mm’ (nn + rr")(nn'+ rr’) from (7) 
 
 Lf Se AO 
 
 =m m'n'n”’.n? + mm” rr? mn’. mm’ ny + m'nr’.m’n’r, 
 
231 
 
 and forming similar expressions for (mm‘n”)* and (rrr?) and 
 adding them together, observing that 
 
 m?>m'm” (n'n” + 9’) = — (mm'm")?, 
 we get 2(mm'm")? + 2(nn'n")? + 2(rr’r’)? 
 pales ” , ” , ry Ay, ray ou , ” Woe 
 =m n'r.m' nr +m’ nro mn'r’ + mn'r m'n"r + mn’. mn’ 
 +m'n'y.mn'r + mn’r.m'nr’ ; 
 and as the second member does not alter when we simul- 
 
 taneously interchange m’ with », m” with r, n” with 7’, the 
 proposed result is established. 
 
 13. To pass from the rectangular system of co-ordinates 
 «, y, x, to another rectangular system a’, y’, 2°; having given 
 the 2@ at which the planes of wy, ay’, are inclined to one 
 another, and the angles @, vy, which their line of intersection 
 makes with the axes of w and w’ respectively. 
 
 Let the new plane of a’y’ intersect the plane of vy in 
 
 the line Ov, (fig. 29), and be situated above it; and let 
 Ox’ perpendicular to the plane a’y’ be the axis of x’, and 
 Ox’, Oy’, the axes of w and y’; and let a sphere be 
 described about O as a center with radius 1, cutting the 
 axes in the points a, y, z, &c.; and suppose these points 
 to be joined by arcs of great circles. Then if va, = ¢, 
 and Za'a,y=0, or xz’ = 0, the plane of a’y’ and the axis 
 Oz’ will be completely determined; also if wa, =, then 
 the positions of Ow’, Oy’, in the plane w,Oy’ will be fixed. 
 
 Also we have 
 , , , , , , 
 C= cosxxty cosyx+2 cosza, &C.; 
 
 and our object is to express the nine cosines in terms of 
 0, p, , which may be done by means of the fundamental 
 theorem in Spherical Trigonometry for finding one side of 
 a triangle in terms of the other two and the included angle. 
 
 First, from the triangle a’a,2, in which Zwv’a,x = 180° ~ 0, 
 
 COS v'&@ = COs @ cos yy — sin @ sin W cos 0 ; 
 
232 
 
 and changing W into 90° + w, 
 cos ya” = — cos d sin Wy — sin d cos yy cos 8 ; 
 also from the triangle z’v,a, in which 2’a, = 90°, za, = 90°— 0, 
 cos *# = sin @ sin 0. 
 Secondly, from the triangle w’x,y, in which a,y = 90° — @, 
 cos wy = cos yy sin @ + sin Wy cos d cos 8 ; 
 and changing y into 90° + Wp, 
 cos y'y = — sin sin @ + cos ycos @ cos 0 ; 
 also from the triangle x’a,y, in which z’a, = 90°, x’ay = 90° + 0, 
 cos ¥y = — cos @ sin 0. 
 Lastly, from the triangle w’v,z, in which za, = 90°, 2’v,z =90° - 0, 
 cos a's = sin Wy sin 6; 
 and changing vy into 90° + W, 
 
 cos y's = cos vy sin @; also cos xs = cos @. 
 
 PROBLEMS ON SeEcrTion I[Y. 
 
 The following Examples will illustrate the method of 
 drawing tangent planes and normals to curve surfaces, and 
 finding their volumes and areas. 
 
 1. To find the equation to the tangent plane to an 
 ellipsoid at a proposed point. 
 2 
 
 ® e @ y* Pod 
 The equation to the surface is — + — + — =1, 
 a Ds eae 
 ve sdz y sds 
 
 we — a tae sey ee O ae 8 oe coe 
 a’ ede °° Bt ce’ dy 
 Webicd dz dz ’ } 
 Hence, substituting for — , —, their values in 
 
 dx” dy 
 
 dz ave, 
 os 4 Tiere ae O — ¥)s 
 
233 
 
 the equation to the plane which touches the ellipsoid at the 
 point vys, Is 
 
 fe + 5% (a0) 45% yy) =o, 
 
 2 , 2 
 so 8 we 
 Cheer apr Raith: Tae Tia eset Lie v= 0, 
 
 c c fi b? b? 
 
 7 
 
 YY 88 
 
 or — + ——+—=1. 
 = b? Cc 
 
 Cor. To find where the tangent plane cuts the axes, 
 ; a 
 making y’ = 0, s’=0, we get w =—, the distance from the 
 @ 
 
 center at which it cuts the axis of #; similarly the distances 
 from the center at which it cuts the axes of y and x, are 
 5? Cc 
 — and — 
 y ra 
 
 2. To find the equations to the normal to an ellipsoid at 
 a proposed point. 
 
 dz ca das cy 
 Since — = —- —-, —= ——%, the equations are 
 
 dx a’ xs dy b? 3” 
 
 ; Cw 
 
 a 2 =[2(@'—2), f y= G2 Ge - 2). 
 
 Cor. By making »’ = 0, we find the co-ordinates of the 
 point where the normal meets the plane of wy, 
 
 , c\ , Gi 
 x“ =x(1-4), Yy -y(1-<)s 
 
 also, the length of the normal, intercepted between the sur- 
 face and the plane of wy, is 
 
 ct a ca? ‘ x 2 
 eI Re eR Ber ea Kae 
 & 
 
 3. To find the locus of the extremity of the perpen- 
 dicular dropped from the center upon the tangent plane to an 
 
 ellipsoid. 
 
 The equations to a perpendicular from the center on the 
 
23 4 
 
 Cw e AREY) 5 J 
 tangent plane are wv = —-—2 = —~ 3", which, combined 
 8 P os ee J Bs ’ ‘ 
 with the equation to the tangent plane, give by eliminating # 
 and y, 
 (a? + y”? +2”) = cs"; 
 ce? 3 aa b?y’ 
 i ye. f eo aengnat 9 a= : — 
 
 in which expressions a’, y’, 2° are the co-ordinates of the 
 point of intersection of the tangent plane and perpendicular ; 
 and w, y, * the co-ordinates of the point in the surface to 
 : au y gt 
 which the tangent plane is applied so that = ae ia 4 mis 1; 
 therefore the required equation is 
 (aa’)? + (by’)? + (c2x')? = (w? + y? + 2)? 
 
 Hence if R be the central distance of a point in the ellipsoid 
 the normal N at which makes angles a, 3, y, with the axes, 
 and P be the perpendicular on the tangent plane at that point, 
 we get from equations (1), considering that P? = v? + y? + 2”, 
 Pcosa=2, a’ cosa = Px, &e. 
 
 fh) J im TLL me die 
 5, 7 =e i eS 
 P as bt t a 
 2 Q 2 
 Gey 
 P? =P? |—+ = +—)=a@ cosa + Bb’ cos’B +c’ cos’ y, 
 ie en I pil hl oe 
 
 R?P? = P’ (a? + y? +2") = a' cos’a + b' cos’ B + c* cos’. 
 
 4. If three tangent planes to an ellipsoid are mutually at 
 right angles, their point of intersection will trace out a sphere 
 concentric with the ellipsoid. 
 
 Let P the perpendicular from the center on a tangent 
 plane make angles a, 3, yy with the axes of a, y, x; then the 
 equation to that plane is a’ cosa + y’cosB + 2’ cosy = P, which 
 must be identical with the equation to the tangent plane 
 
 expressed by the co-ordinates of the point of contact, viz. 
 
 ve yy 82 
 Pnerre ean e o 
 2 
 
 a” b 
 
235 
 
 xv P yP xP wana. «38 
 *. cosa = cos 3 =——, cosy = —, but —~+—4-—=1 
 a a. > B i? 2) Py Cc ? a b° Cc ) 
 a P\* y P\? 3P\? P 
 . Pia ( oo (=) 4s (==) =a’cos’a + b* cos’ 3 + ¢ cos? +y 
 a Cc 
 
 as before, which gives the length of the perpendicular in terms 
 of its inclinations to the axes. Now let there be two other 
 perpendiculars P’, P’, which make with the axes the angles a’, 
 
 , , 
 
 , ° 
 B's a”, B’, ys respectively, 
 “. P®? = a’ cos’a’ + b° cos’ 3’ + & cos’y’, 
 P’”? = a? cosa’ + b’ cos? 3” + ¢* cos’ry’” ; 
 
 and suppose the three perpendiculars to be mutually at right 
 angles; then since a, a’, a’ are the angles which a line (viz. 
 the axis of v) makes with three rectangular axes, viz. the 
 three perpendiculars, if we take account of the condition to 
 which they, as well as B, 8’; B’, y, y'5 y's are subject (Art. 
 5), and add, we find 
 
 (Pete Pcs Pie Gay +e 
 
 but the first member is equal to the square of the distance 
 from the origin of the point of intersection of the three 
 planes; this point therefore describes a spherical surface con- 
 
 centric with the ellipsoid whose radius = J a + b? +c’. 
 
 In the case of hyperboloids, one at least of the quantities 
 a’, b’, ce’ is negative, and hence their sum may be negative or 
 nothing ; in the former case there is no point in space through 
 which three rectangular planes touching the hyperboloid can 
 be drawn, and in the latter, the center is the only point which 
 has that property. 
 
 Cor. If we remove the origin to the extremity of the 
 semiaxis a, and then make a infinite as in Art. 72, the equation 
 to the ellipsoid will become 
 
236 
 
 and the equation to the sphere will become 
 oe —2an+y +2" =2ap — p> +2ap — p”, 
 
 or, making a infinite, - # = p +p’; therefore the locus, in the 
 case of paraboloids, is a plane perpendicular to the axis of the 
 surface, and at a distance = — (p +p’) from the vertex. 
 
 5. Three planes mutually at right angles constantly 
 touch the perimeter of a plane curve of the second order ; to 
 find the locus of their point of intersection. 
 
 Let the equation to the curve, and to one of the rect- 
 angular planes, be respectively 
 Li beet of; : : ; 
 atiani, acosaty cos3 + % cosy =d; 
 then making x’ = 0, the equation to the trace of the plane on 
 wy is x cosa + y' cos3 = d, which must be identical with the 
 equation to the tangent of the curve 
 
 U , 
 CR yy vd yd 
 Sy yer Boel that cosa = GP? cos} = ra 
 
 2 
 
 way Ned ate 
 = =") an (4) = a° cos’a + 6’ cos?3; and similarly 
 
 , © id 
 d® =a’ cos*a + 6 cos*', d’? = a’ cos’a” + b’ cos’. 
 
 Hence, adding, d?+ d? +d”? = a? + b*, which shews that 
 the locus is a sphere, concentric with the curve, whose ra- 
 
 dius = \/a? + b°. 
 
 Cor. If the curve be a hyperbola, the radius = J a — 
 therefore we must have the transverse greater than the con- 
 jugate axis, otherwise the problem is impossible; if a = 6, 
 asin the rectangular hyperbola, the locus is a point, viz. the 
 center. If the curve be a parabola, it may be shewn, as in 
 Prob. 4, that the locus is a plane perpendicular to the plane 
 of the parabola passing through its directrix. 
 
 6. If two concentric surfaces of the second order have 
 
237 : 
 the same foci for their principal sections, they will cut one 
 another every where at right angles. 
 x’ y 3? £ y 3 
 Behr ag chs bing sails oe y Ga = 1, 
 be the equations to the surfaces; then the equations to the 
 tangent planes are 
 
 Po 
 nx ie! y cena? © YY Se 
 
 ‘ b) Sha. F Tiare 3 
 a? vege a’? b? ce? ‘ 
 
 and in order that these may be at right angles, we must have 
 
 Ga?” BBP ee? 
 (hence one of the surfaces must be a hyperboloid, for some of 
 the quantities a’, b*, &c. must be negative;) and this equation 
 gives the relation among the co-ordinates of the points in 
 which the surfaces may intersect at right angles. But by 
 subtracting the two equations, we find 
 
 1 1 3 1 1 4 1 1 , 
 aba as) ge OND Abbi F 2 Netacok eames 
 
 for the relation among the co-ordinates of the actual points 
 of intersection, which must be identical with the former if 
 the surfaces intersect every where at right angles. Hence, 
 equating the ratios of corresponding coefficients, 
 
 a—a’?=? —-b? = — ec”, 
 6 , , iD , © a 
 or @ —- P=a’*-b*, @-eC=aa*?—c*, Yb -—e’=b? —c?, 
 
 which three equations express that the principal sections of the 
 surfaces have the same foci, or that the surfaces are homofocal. 
 
 Cor. In like manner, if the surfaces have not a center, 
 and we represent their equations by 
 
 2 
 
 Y 
 Ir la ars 
 b*agiia 
 
 i Za 3° 
 (Sat aa ihe 
 
 by eliminating w between these equations, and comparing the 
 
238 
 
 result with the equation expressing that the tangent planes are 
 at right angles, we find that the surfaces will intersect every 
 where at right angles, provided 
 
 BF 0 ae ee — 2p” 
 
 12 
 
 CC — aes 
 
 2a Pig at 2a 2a) 
 
 that is, the foci of the principal sections of the surfaces must 
 be coincident. 
 
 7. If three surfaces of the second order have the same 
 foci for their principal sections, and if each of three planes 
 mutually at right angles touch one of them, the point of inter- 
 section of the planes will trace out a sphere. 
 
 8. If the tangent planes to two homofocal surfaces of 
 the second order be parallel, the difference of the squares of 
 the distances of those planes from the center of the surfaces 
 is invariable. 
 
 9. The locus of the projection of the origin of co- 
 ordinates upon the tangent planes to wyx = a*, has for its 
 equation 27a°wyz = (a + y’ + 2°)’. 
 
 10. If the equation to a surface be \/v+\/y+\/ 2=\/a, 
 the sum of the portions of the co-ordinate axes, intercepted 
 between the origin and tangent plane at any point, is constant, 
 
 11. Ifwyszbe the point in a curve surface the tangent 
 plane at which forms with the co-ordinate planes a pyramid 
 of the least volume, the equation to that plane will be 
 
 12. The number of normals to a surface of the n‘" order 
 that may pass through a given point cannot in general exceed 
 V—-n +n. 
 
 Take the given point for the origin, and suppose the 
 equation to the surface to be 
 
 w= 3" + Vis"! 4 Vis"? + ...4+ 0,124 V, =0, 
 
239 
 
 where V, denotes an integral function of w and y of r 
 dimensions; then 
 du du R+@Q ie 
 RS a5 ee © =0, or +Qp=0; 
 similarly, S + Qq = 0; 
 
 consequently the points of the surface through which the 
 normal drawn from the origin may pass, will be determined 
 by the values of w, y and x that simultaneously satisfy the 
 three equations 
 
 u= 0, Qe -Rz=0, Qy-—Szx=0; (1) 
 
 and as each of these equations is of m dimensions, they will 
 determine * points. But if x =0, w and Q are reduced to 
 V, and V,_,; and equations (1) are replaced by 
 
 %=0, y= Very 0s 
 
 which determine m (m — 1) points in the plane of vy that may 
 be foreign to the question, and such (not that the normal at 
 them passes through the origin but) that at them the tangent 
 plane is parallel to the axis of x; consequently the number 
 of points determined by equations (1) that satisfy the proposed 
 condition cannot in general exceed n’?— ”? +n. 
 
 13. If p, 7 be the perpendicular on the tangent plane 
 and the radius vector at any point of a surface, then p* =r 
 will equal the length of the perpendicular on the tangent plane 
 at the corresponding point of the surface which is the locus of 
 the extremity of p. 
 
 Let wv, y, be the co-ordinates of a point in the first 
 surface, a, 3, yy those of the corresponding point in the second 
 surface; then we have the relation 
 
 av+Bytys=p=ae+h?+y’, (1) 
 
 since a, 3, y are the co-ordinates of the extremity of p. But 
 this being the equation to the tangent plane at (wyz), the co- 
 ordinates w, y, may receive indefinitely small variations 
 
240 
 
 without the magnitude or position of p being altered; there- 
 fore we may differentiate considering a, 3, -y as constant, 
 
 . adw+ Bdyt+ydzx=0. (2). 
 
 Now if V = f(a, B, y) = 0 be the equation to the surface 
 traced out by the extremity of p, and P the perpendicular on 
 the tangent pas to that surface, 
 
 poate Bagt va, +N (a) + Ga) + Ge): 
 
 and from the oats to fe surface 
 
 ond +o 5 dB + dy =0 (3) 
 
 But differentiating (1) pa it and taking account of 
 (2) we get 
 (w - 2a)da+ (y-—2B) dB + (z - 2y) dy =0; 
 
 hence comparing the two latter equations we have 
 
 d 
 Therefore substituting for ae &c. in the value of P, and 
 ‘A 
 
 reducing by means of equation (1) we get 
 
 _a(w7—2a)+BYy-2B)+y(8@-2y) P+ PhP +y Sa 
 
 J (@—2a)* + (y— 23)? + (s-2y)? "a +y°42? 7 
 
 Application of the Formule for the Volumes and Areas of 
 Surfaces to Examples. 
 
 1. To find the volume and area of the surface of a 
 sphere, (fig. 41). Here 
 
 av 
 
 — — SL ae talers 
 dudy od / a v Y's 
 
241 
 
 + C, 
 
 dV é 
 . —-=taVYa-y-wv#+4(ae-y’) sin 
 a” — y° 
 
 and integrating between the limits 7 =0, r= MQ= /a — ¥°, 
 
 *, integrating between the limits y = 0, y= CM, 
 volume AMBQ = (a’y —4y’), 
 
 3 
 and, making y = a, vol. ABCD = = = = ae 
 
 ae Andra 
 *. vol. of whole sphere = 8 a= sos 2 
 
 as as\4 dz\? 
 Ranger a, (<* ca 
 oe dady / i e %3 i 
 
 ds 
 oe aot ib +C; 
 dy / a? Le y? 
 and taking the integral between the limits #7 = 0, v=\/ a?-y’, 
 dS 1wa_ 
 dy 2” 
 
 .. integrating between the limits y = 0, y= CM, 
 area of surface PBAQ = 7, 
 ; Ta 
 and making y = a, area of surf. DAB = ai 
 
 2 
 7 a 
 ‘, area of surface of whole sphere = 8 ers 47a’. 
 
 16 
 
242 
 
 2. The volume of the solid which is common to two 
 equal spheres that have their centers in one another's surfaces, 
 5ara 
 12 
 
 1S 
 
 8. To find the area of the spherical surface intercepted 
 by a right cone whose vertex is in the center of the sphere, 
 and whose vertical angle is 2a. 
 
 ds > @ 
 As in Ex. 1, — = asin7=1———— +0, 
 dy 
 
 fay 
 
 and integrating from # =0 to w= / a? sina — y’, 
 
 ds Vie asin’ a — y” 
 —— = asin~ ag 
 y 
 
 a —y? 
 
 and again integrating from y=0 to y=asina (Integ. Cal. 
 Art. 89), 
 
 9 
 
 4 
 
 a 
 S= a (1 — cos a). 
 
 4. To find the area of the spherical surface intercepted 
 between two meridians inclined at an angle @, and a plane 
 perpendicular to one of them at a distance c from the center, 
 and parallel to their line of intersection. | 
 
 dS : 
 = Pte See Bt Cc 
 dx J a? — x 
 : wv tan @ 
 =a sin“} , from y=0 to y=a@£ tan@; 
 
 \/ a — 2 
 x tan @ v 
 
 = - @ tan @ f ——______ 
 Va wv # (a? — x) \/a? — x sec? 0 
 
 * §'=awsin—} 
 
 xv tan@ een es 
 f SIN Ge eee OO, 
 
 Va — a / a? — x? 
 
 =aasin=! 
 
243 
 
 and integrating from #=c to w=a, we get the surface 
 required, 
 
 Pie we OSI 7 : c tan @ 
 2S = 2a" sin~* ————— _— 2acsin™ 
 
 en 
 J/ ae aA ile? 
 
 5. To find the volume intercepted between the surfaces 
 whose equations are 
 
 Party, e=ar+y’ 
 
 dxdy 
 dV ——__— yt /e+ ye 
 Ten dy Vara + hat log (VEVE EY) 
 
 and taking the integral from y=0 to y =\/a? — a’, 
 
 9 
 
 dV = 3 a a — a 
 Lan/@ — a + yeaa 
 av 
 
 therefore integrating between the limits vy = 0, w =a, we get 
 
 (Integ. Calc. Ex. 5, p. 97) 
 
 Te ra ra 
 
 le ail See la ae 
 ey rye 
 
 Aa a 
 
 *. volume required = 8V = 
 
 6. The axes of two equal cylinders intersect at right 
 angles, to find the volume and superficial area of the portion 
 which is common to both. 
 
 Let OC (fig. 46) be the axis of one cylinder, O.4 that 
 of the other; then the equation to AB or to the cylinder 
 
 whose base it forms is s = 4/ a? — y’, 
 
 av dV 
 =<\/a’>—y’, and [ma Bais nk sal 8 
 
 ia dady — 
 
 16—2 
 
24:4 
 
 integrating between the limits 7=0, w=ON= J a -y’; 
 hence, integrating again between the limits y =0, y=a 
 
 ae 16a3 
 vol. ADMB = oe and whole vol. intercepted = 
 dz -y dz ad’ S a 
 A 1 ——— ee | ee “gee 0, one = —_}~ ——_ § 
 Bane? dy \/ a? a y : daw dady (yg y 
 - ax 
 
 + C=a from #=0 to va/fa-y <n 
 at é 
 
 hence area of ADPB = a’ = area of CDPB, and area of whole 
 surf. = 16a’. 
 
 7. If the axes of the equal cylinders intersect at an angle 
 
 16a? 
 
 a, it will be found that the volume common to both is 
 
 8sina’ 
 
 16a’ 
 
 and the surface 
 
 8. The volume of the figure generated by a straight line 
 which moves parallel to the plane of yx and constantly passes 
 through the axis of # and through the perimeter of a circle 
 whose plane is perpendicular to the axis of y and center in 
 that axis, is 2 1 za’b, a being the radius of the circle and b 
 the distance of its center from the origin. 
 
 9. To find the volume contained between the co-ordinate 
 planes and the surface 
 
 we NEE Nae 
 
 The section of this surface by a plane parallel to wy at F 
 
 2 
 distance x’ from it, is the parabola vee J! =l]-— Wee 
 c 
 
 and the area contained between this curve and its two tangents 
 
245 
 
 ab 
 6 
 
 — 
 — 
 
 3 » e e . 
 (1 - /*) ; consequently integrating this from z’=0 
 Cc 
 , “ 1 
 to x = c, we get the required volume = iG abe. 
 
 ; x y° 3" 
 10. The volume included by the surface act ra +—=1 
 a . a 
 
 2n 
 
 is arabe. 
 
 11. The volume intercepted in the positive compartment 
 
 c! 
 
 of vy by the surfaces vy = az, 2 +y’ =", is }—; and the 
 
 a 
 
 area of the intercepted part of the surface vy = az is 
 
 Ta ce?) 3 
 erat: e fereahe| eT 
 6 ( +5) 
 12. To find the differential coefficients of the volume 
 and surface of a solid, bounded by two of the co-ordinate 
 planes and a plane parallel to the third, and by a plane 
 
 through one of the co-ordinate axes, and by a given curve 
 surface. 
 
 Let DMQ (fig. 40) be a curve surface, QD being its 
 trace in the plane of za. Draw planes MP, Np parallel 
 to yx, and MA, NA planes passing through Az. Let 
 a, y, x be co-ordinates of the point W and ¢ =tan z HAP, 
 therefore y= wt; also let r+h, (w~+h) (t+ e) be co-ordi- 
 nates of the point K. Then AAHP = 2x’ t, 
 
 h 
 and area KH = (wx + hy - 2° =he («+ | : 
 
 Now the volume DMP may be considered as a function 
 of # and ¢; and by subtracting the increments arising from 
 their separate variations from that arising from their simul- 
 taneous variations, it may be shewn that 
 
246 
 
 hence, taking the limit of both sides, by making hf and e 
 vanish, in which case vol. MKmk, becoming ultimately a 
 prism whose base is HK and altitude MH, =xhe (a+ 5h), 
 we have 
 a Vi 
 | dn dat 
 Also, if S’ = area of the surface DMQ, by a similar process 
 we may shew that 
 surf. MN ds 
 “ he © dedt 
 
 and taking the limit of both sides, in which case 
 
 UR 
 
 + &c.; 
 
 h 
 surf. MN = HK secry = he (w+ 3) sec 4, 
 
 dS Ne —tds\? Gy 
 i Vv <= = WW —— — - 
 we have HE @ SEC ty Vv + ( ) + 
 
 dx dy 
 
 In applying these formule to examples, the substitution 
 of xt for y gives 
 av ds 
 
 ei YL DAS (yy Fee 
 Tite Me EES 
 
 integrating with respect to «#, considering ¢ as constant, 
 between the limits 2 =GN=¢(¢), «= GN’=\, (t) (these 
 values being obtained from the equations to the bounding 
 cylinders AN, A’N’) (fig. 42) we obtain quantities, functions — 
 of ¢ only, which when multiplied by d¢ are respectively the 
 ultimate values of the volume and surface of the wedge 
 PM’; and these integrated between the limits =m, t=m’, 
 will give the volume and surface contained between two 
 planes inclined to zw at angles whose tangents are respec- 
 tively equal to m and m’. 
 
 EOE (2, 07) 
 
 13. <A sphere is pierced by a cylinder the diameter of 
 whose base is equal to the radius of the sphere, and their 
 surfaces are in contact; to find the volume and area of 
 the surface of that part of the sphere which is intercepted 
 by the cylinder. 
 
2447 
 
 Taking the center of the sphere for origin, and planes 
 drawn through the axis of the cylinder and perpendicular to it 
 for those of xv and wy, and using the formula of Prob. 12, we 
 have 
 
 RV es ae Bie. 
 =0s=a\/a—-—-v -—y=v7V/a-—-(1+2%)2'; 
 Pee J y V/ lier ts) 
 3 3 
 peSlimog, 90 Bie 2 \etea tau foo 
 dt 714+¢ Sll+ 2 aa epi? 
 
 a 
 from v= 0 to x = AN = acos* MAN = siete (fig. 44) ; 
 
 2 
 V at yrds : egies @ 
 ie = —< tans a ob 
 3 V4 Pat? 7 a+e) ¢ 
 
 and integrating between the limits ¢=0, t= 0, 
 
 3 3 
 vol. ABPM =—{" 2} =e aaa 
 
 Now vol. ABCD ="— ; therefore the part of ABCD not 
 
 9g 3 
 comprized in the cylinder =—— . Consequently, if the sphere 
 
 be pierced by two equal cylinders, the part not comprized in 
 
 2 
 them = Aaa 
 
 Again, for the area of the surface 
 
 as Ry: dz\? 
 dx 
 
 dx dt 
 
 (j=) <« Cage nk IE Sa 
 dy a 5 Ve a(t 2). 
 lel arte Vea FOFF) = 0 : -ami 
 
 dt Tee 1+ (1+#)8 
 
 taking the integral between the limits # = 0, # = ra as 
 cs 
 
 .S= a? (tan4 + 5 os 
 
 + C, 
 /1 + =) + 
 
248 
 and taking the integral between the limits ¢=0,f= 0, 
 
 area of surf. BDP =a (z _ 1) 
 
 2 
 Now area of surf. BCD = a therefore part of BCD not 
 
 intercepted = a?; and if the whole sphere be penetrated by 
 two equal cylinders, the area of the surface not comprized 
 in them = 8a? = 2 (diameter)’. 
 
 14, To find the volume of any wedge of an ellipsoid 
 intercepted between two planes passing through an axis. 
 
 (fig. 41) 
 
 av / oY? wi MT lineat® 
 ——_ = &@2 = C# se aoa ae 1-—-@ Gaace 
 
 dadt a 
 dV betes ad he ie okG 
 BP Whi Pree “(5+5) ioe ee 
 a BP oe 
 1 
 integrating between the limits 7=0, v= MQ = 
 lea’: 
 a Be 
 foram the ploncice ees a eee 
 or in the plane of wy, — saag8 em at ait paimaas 
 
 be t 
 = —— tan? a = vol. of wedge ABQC, 
 
 b 
 and making ¢ infinite, vol. ABDC = =. ze 
 Ea yol fof whnlenel i neord ease 
 
 15. To find the volume of a spherical sector which is 
 inclosed by the spherical surface, and a conical surface having 
 the center of the sphere for its vertex and one of the circles of 
 the sphere for its base. 
 
249 
 
 Let a be the radius of the sphere, and 2a the vertical 
 angle of the cone; then by the formula of (Art. 118), 
 
 3.3 dV 3 3 
 SP Woe NBL cow 
 
 dddp 8 ° d¢ 
 between the limits @=0, 0=a; therefore integrating again 
 between the limits ¢ = 0, @ = 27, 
 Q7a° 
 volume of sector = ee (1 — cosa). 
 
 16. The volume contained between the co-ordinate planes 
 and the surface which is the locus of the projection of the 
 
 origin on the tangent planes of wyx = a*, may be shewn by 
 3 
 
 9a 
 polar co-ordinates to equal tk 
 
 17. To find the area of the surface of an ellipsoid. 
 
 Let DR (fig. 41) be a section parallel to the plane of 
 vy at a distance =ccos@ from it, and BR a section through 
 the axis of x, inclined to the plane of s# at an ZW, and let 
 
 tan = = tan @; and let S = area of surface BDR; then S is 
 
 a function of @ and @, and if DR’, BR’ be sections cor- 
 responding to the angles 0 + 60 and @ + d¢, we shall have 
 ad’ § ees aa? of surf. RR’ 
 dpd0 dopo 
 Now if OR =p, Or = p+ Ops and ry = inclination of tan- 
 gent plane at R to wy we have ultimately 
 
 area RR’ =(its projection on wy) sec y = pop ow sec x. 
 Q 
 
 . ee 3 ' 
 But the equation to RD is 2 + a = sin’@, 
 
 ? cos? ? sin?! ? cos? : 
 al a + —— = core (1 + tan? ~) = sin’, 
 
 Layer a ae and Gye eon Gee 88 : 
 
 cos Vi cos Vy 
 
ay od; ultimately ; 
 
 , area RR’ = ab sin O cos OS Ped sec y; 
 
 eer 
 
 + aon" ab sin@cos@ — Nf, - ie 
 =absin@\/1— Su. cos’ @ + y sin’ p} sin? @, 
 
 a— Cc 5? — ¢? 
 making a gli a sean 
 
 and observing that w= pcos\=asin@cos¢, andy =bsin@sing; 
 this can be integrated only by elliptic transcendants and gives 
 the whole area 
 
 Q2ab 
 
 ae ge ee ee I cp ete a 
 Ferg leF@+@ - EC} 
 CAML 
 where a cos a = ¢, and the square of the modulus = est) . 
 b* (a — c*) 
 
 If the ellipsoid differ little from a sphere, making 
 
 p. (cos ~)” + v (sin P)’ = uv 
 
 (which is a small quantity in the case proposed) and expanding, 
 
 ai ab sin 6 $1 — Lu (sin 6)’ HE? (ai 6)*-& 
 = — — —— n = K, 
 dpdé 2 2.4 re 
 .”. integrating between the limits 9 =0, 9=4z, (Int. Cal. p. 126) 
 ds 1 1 
 FES en So Tae oR ete 
 and integrating again between the limits 6=0, P=47, 
 
 we find the area of the surface of dth of the ellipsoid 
 
 aw ab 1 1 
 ae me 2a es ies 4. 
 Gee 3 ihesjl I Rees: 
 where P}= 3 (uty), Po =p? Ft ae ——-; &c. 
 
251 
 
 and in general P,, is the coefficient of "in the developement 
 
 of (1 — wx)72. (1 — ve)73. 
 
 18. ‘Lo find the area of the surface of an oblique cone on 
 an elliptic base. 
 
 Let the perpendicular VW (fig. 47) upon the base from 
 the vertex of the cone fall upon CA the major axis of the 
 ellipse, CM =f, VM=h; w«=asingd, y = bcos @, co-ordi- 
 nates of P, BP=s, VPT the tangent plane along VP, 
 cutting VMP’ in VQ, VH a perpendicular upon PT, 
 S'= area of surf. BVP; then ultimately 
 
 =e op = surf. VPP’ = triangle VPQ = 4PQ.\VH=1VH.0s; 
 
 s CITE: YH L/h (a? — cc’ sin’d) + 0° (a —fsin d)’, 
 as will be found upon calculating VA, since © = / a —c'sin’ —c’sin <0 
 
 where c’ = a’— b*; this can only be eral by elliptic func- 
 tions, unless 2 and f be so related that b°f?=c’(h? +6?) when 
 the quantity under the vinculum is a perfect square, and 
 
 awabf 
 ins 
 
 whole surface of cone = 
 
 19. To find the whole area of the surface which fs the 
 locus of the projection of the center of an ellipsoid on its 
 tangent planes. 
 
 Employing polar co-ordinates we have (Art. 120) 
 
 a’ 3 aq J ; ae : 
 ee = a ry sin@, since == Pe ELOumiam). 255), 
 
 where 7” is the radius vector of the ellipsoid at a point where 
 the normal makes an angle @with the axis of zs and its 
 projection on the plane of wy makes an angle @ with the axis 
 of w, and r is the perpendicular on the tangent plane to the 
 
252 
 
 ellipsoid at that point; consequently by Prob. 3, p. 234, we have 
 a’ s 
 
 TVR 
 
 = sin 9 /(a’ sin 0 cos p)* + (B’sin 6 sin )” + (c’ cos 6)”. 
 
 But for the surface of an ellipsoid with semiaxes a’, b’, c’ 
 we have (Prob. 17), 
 
 as 
 dodo 
 
 which coincides with the preceding expression if 
 
 =sin 0\/ (a’b’ cos 0)” + (b'c' cos @ sin 0)’ + (a’c’sin @ sin@)? 
 
 , ‘ 
 ab=c’, be =a’, ac =0'; 
 
 on Cla sO Clee eeL 
 
 or a4 = “eas ee Cc = - ° 
 a b Cc 
 Consequently, as the limits of the integrals are the same, the 
 area of the surface which is the locus of the projection of the 
 center of an ellipsoid (with semiaxes a, 6, c) upon its tangent 
 planes, is equal to the area of the surface of another ellipsoid 
 ac ab 
 
 : ; be 
 with semiaxes —, —, —. 
 = UaeG 
 
 ProspiEMs oN SEcTION V. 
 
 The following are examples of finding the equations to 
 curves in space, and drawing tangents, normal planes, &c. 
 
 1. A sphere is pierced by a cylinder the diameter of 
 whose base is equal to the radius of the sphere, and their 
 surfaces are in contact; to find the nature of the curve formed 
 by their intersection. 
 
 Take the center of the sphere for the origin, and planes 
 drawn through the axis of the cylinder and perpendicular to 
 it, for those of zx and wy respectively; and let 4N =a, 
 NM =y, MP = (fig. 44) be the co-ordinates of a point P in 
 the curve; and dB =2a. Then because P is in the spherical 
 
253 
 
 surface, vw + y® + 2” = 4a°, and because M is a point in the 
 
 semicircle a + y° =2aax; therefore the equations to the pro- 
 
 jections on wy and sa, or the equations to the curve, are 
 y=2ax—- 0°, 8° = 4a" -2aa...(1); 
 
 the latter representing a parabola BD, vertex B, and axis 
 AB; if we eliminate wz, we find 4a’y’ = 4a’z” — x‘ for the 
 equation to the third projection on yx. 
 
 Differentiating equations (1), we find 
 
 dy a-@ d’y a ods a d’ x a® 
 dx Tee. dat ens eda" 3 | 
 
 Hence the equations to the tangent P7' at a point wysx, are 
 
 a— 
 
 Yen — 
 
 wv a 
 (a'-@), x -g=-—-(a'-2); 
 % 
 
 and the co-ordinates of its trace 7’ on the plane of wy, making 
 % = 0, are 
 3 
 ? , 
 e©=404—-% Y= 7 
 
 The equation to the normal plane is 
 v—x@2+ rapa! Yy) -=(s/-2) = 0; 
 and the equation to the osculating plane 
 (y —y)y — (e — x) x= (a — a) fax’ + (a a) yf. 
 Also if s = length of arc DP, 
 
 toe wie ie a) (=) -! Ee) ian 
 
 da Q2v 2a—2 
 
 let w = 2acos'@, .. —_ Dp) are: — $sin’d 
 
 therefore, integrating between the limits 0 =i7, 0=0, 
 
 arc DPB = elliptic quadrant semiaxes 2a WA 2, and 2a 
 
 2. To find the equations to the Helix. 
 
254 
 
 Whilst the rectangle ABCM (fig. 45) revolves uniformly 
 about its side AB, the point P moves uniformly along the 
 parallel side MC, and generates a curve called a helix. 
 
 When the rectangle is in the plane of zw, let P be at 
 M, and let the velocity of P= times the velocity of M, 
 “*. PM=n.arc DM; also lett dAN=a, NM=y, MP=s 
 be co-ordinates of P, and 4M =a, 
 
 xv es 
 - g=mnacos-!—, and y=\/a@_-a’; 
 a 
 
 which are the required equations; the former may also be 
 
 written ¢ = ma sin! J 
 a 
 
 Cor. Since the corresponding increments of DM and 
 PM are always in a given ratio, the curve DP always cuts 
 the generating line CW of the cylinder on which it is traced 
 at the same angle; and the line touching DP is always inclined 
 to the plane of wy at a constant angle a whose tangent = 7 ; 
 therefore also the length of the helix 
 
 x 
 DP =seca.arc DM =asecacos™!-. 
 a 
 
 3. Let A (fig. 36) be the pole of the great circle «By 
 of a sphere whose center is C; and as the quadrant AB 
 revolves uniformly about AC, let the point P move uniformly 
 along it; to find the equations to the locus of P. 
 
 When the plane ACB coincides with za, let P be at A, 
 and let B’s velocity = times that of the point P, 
 
 - arc Ba=n.arc AP; 
 also let CN=a, NM=y, MP=2s, be co-ordinates of the 
 point P, and AC =a, 
 
 | g 
 
 - tan? 2 = ncos7 -, and 2? +y?+2° =a’; 
 v a 
 
 which are the required equations. If m=1, these equations 
 may be transformed into 
 
 av=x /a— 2°, ay = a’ — 2’, 
 
255 
 
 Cor. To find the area of the spherical surface included 
 between the co-ordinate planes and the path of P. 
 
 Let AB’ be a position of the quadrant very near to AB, 
 meeting the curve in P’, and a section of the surface through 
 P perpendicular to AC in p; then if ACP=0, ACP’=0 +60, 
 we have the increment of the area dw BP = PP BB’ = PpB'B 
 ultimately ; hence (Int. Cal. p. 182.) 
 
 oS = BB’.PM=n.P'p.PM=n.a00.acos0, ultimately, 
 or on = na’ cos 0. 
 
 Therefore area of surface 4a BP = na’ sing. 
 
 This result may of course be obtained by means of the 
 formula of Art. 113. 
 
 Also if V be the volume contained between the co-ordinate 
 planes and a conical surface whose directrix is the path of P, 
 
 we have, making 4 BC# = qd, 
 
 d2V 
 =a sin@; 
 dpdé 
 dV 3 3 
 °, Ee 6 Rec) = 4 cosh: from gouCP Shh to@=90°: 
 da 3 3 n n 
 ; 3 
 ee a from @ = 0 to @ = BCa, 
 5 n 
 or volume Bed Ce sin 0, 
 
 3 
 
 4. To determine the curve to be traced on the surface 
 of a sphere so that its length shall always be equal to m times 
 the ordinate x of the describing point. 
 
 Let x, y, x be the rectangular co-ordinates of the de- 
 scribing point, and r, @, @ the polar co-ordinates, then we 
 are required to assign a relation between @ and @ which will 
 
 2 
 1 a. a Ss 2. 2 e 20 ® ° t t 5 
 give s = ”2Yr COS GU, OF do =nr sin’G, since 7” 1S constant 5 
 
 therefore (Art. 132) 
 
256 
 
 } dq * dp eee 
 ] 2 a B. a 2 —_—- «= 2 —_ 
 + sin @ ( ) m sin’@ or amass nen a8 sin*@? — 1; 
 
 .O@+C= eo + sin7* Ga (Integ. Cal. p. 120), 
 n— 
 
 J +C=ncos™! +sin-! 
 
 NB x 
 VA eee r/n?— J 2 1/7 ~ 2 
 the equation to the projection on the plane of yx; the 
 constant C' may be determined if any point be given through 
 which the curve is to pass. 
 
 or sin~! 
 
 5. To find the line of greatest inclination at any 
 point of a curve surface; that is, among all the straight 
 lines which touch the surface, to determine that whose in- 
 clination to the plane of wy is the greatest. 
 
 As all these lines must lie in the tangent plane at the 
 proposed point, the line required is obviously that which is 
 perpendicular to the trace of the tangent plane on wy. Now 
 the equation to that trace is 
 
 , Pro ad ’ ! Pe 
 —-y=-—--(w-a)--3; “yY -y=-(e-@ 
 y F i y at ) 
 is the equation to a line perpendicular to it, passing 
 through the point (a, y), and is therefore the equation to 
 the projection on the plane of wy of the line required ; 
 the other equation 
 
 v-e= A gst (2’ — x) 
 
 is obtained by combining the former with the equation to 
 the tangent plane (in which the line in question lies) so as 
 to eliminate y/ — y. 
 
 6. Hence we can determine the curve of greatest in- 
 clination on any surface, that is, one whose tangent is always 
 inclined at the greatest angle to the plane of vy. 
 
 Suppose y = (wv) the equation to the projection of the 
 
O54 
 
 curve, then the equation to the line touching the projection at 
 a point (a, y) (which must be identical with the equation to 
 the projection of the line of greatest inclination) is 
 
 ’ dy, d 
 goers Hed 
 
 “da p 
 
 If, therefore, we determine p and g from the equation to 
 the surface, independent of x, and substitute them in the 
 above equation and integrate, the result will represent a 
 curve on which if a cylinder be erected, it will intersect 
 the surface in a curve having the property that its tangent 
 is always inclined at the greatest angle to the plane of vy. 
 
 Ex. Let the surface be of the second order represented 
 by the equation 
 
 A & By 
 A B ; 2 D oe ect hg perp a ar ie $ 
 ities: ae & ? Exige ©GR 
 B e , 
 os = = a or, integrating, y4 = Cx. 
 wv v 
 
 Hence the curves of greatest inclination are of double cur- 
 vature, and are projected horizontally into parabolic curves 
 if 4 and B have the same sign; if, however, the point of 
 departure, which determines the constant C’, be in a prin- 
 cipal plane, z# for instance, we have y=0 when @@=@, 
 and therefore C’=0; hence the curve of greatest inclina- 
 tion is the principal section in zw since its equation is 
 reduced to y = 0. 
 
 7. On a given surface to determine the curve of con- 
 stant inclination, that is, one whose tangent line is always 
 inclined at a constant angle to the plane of vy. 
 
 Let x = f(a, y) be the equation to the surface, p and q 
 the partial differential coefficients of x; also suppose y = ¢ (a) 
 to be the equation to the projection of the curve of constant 
 inclination, and yy the angle at which the tangent line at a 
 point wyx is inclined to the axis of x; then (Art. 131) 
 
 ds dz i (= 
 
 2 Ser te =) 
 OT dae da’ aa Tpuata sp Ge 
 17 
 
258 
 
 Be oy Gaal la. 
 or + {——]) = tan —};3 
 dx Y \de 
 but, vw, y, x being co-ordinates of a point in the curve, we 
 
 have z = f(#, y) with the condition y = f(a) ; 
 dz d 
 
 dx da 
 
 dy Wh dy\? 
 ane t * aad = 1 a 3 
 hey (p Ht =] fj (5*) 
 
 and it remains to substitute for p and q their values in terms 
 of wand y, and to integrate this equation. We shall then 
 have the required relation y = @ (a). 
 
 d dy dy 
 Sf (@%y) ti ae) sia? hd aye 
 
 8. When the given surface is one of revolution, the 
 equation to the projection of the curve of constant inclination 
 on a plane perpendicular to the axis, may be conveniently 
 expressed by polar co-ordinates ¢, p, so that | 
 
 w= pcosd, y=psin d. 
 
 2 2 
 As above we have 1 + (5”) = tan’ cy : 
 \aAP xv 
 
 dd\? ds\? 
 “. 1+ 9° a = tan’. eA : 
 
 2 2 aot? 
 which may also be put under the form (") ag la 2 
 dp} p’—p 
 
 p being the perpendicular from the origin upon the tangent 
 to the required projection. Now from the equation to the 
 
 : dz. 
 generating curve x = f(o), we can find 4. in terms of p, and 
 hence obtain the differential equation to the required projection. 
 
 Ex. Let the surface be a paraboloid, and the origin 
 in the center of its base, ce its altitude, @ its semi-latus 
 rectum 5 
 
 “. (€— 2) 2a= 9’, AE 
 
259 
 
 or, p’— p’ = a’ cot’ y, 
 the equation to the involute of a circle whose radius = a cot ry. 
 
 If the surface be a right cone, the projection is an 
 equiangular spiral. 
 
 9. To determine a curve traced on a surface of revolu- 
 tion, which always cuts the generating curve at the same 
 angle. 
 
 Let DP (fig. 48) be the curve, EN its projection on 
 a plane perpendicular to the axis of the surface; CPG 
 a section of the surface through its axis meeting the curve 
 in P, PN=s, DP=s, ON=p £AON=d®, and a the 
 constant angle CPD at which the curve cuts the meridian ; 
 also let CQP’, a meridian plane very near to CPG and 
 inclined to zw at an angle @ + od; cut the curve in P’, 
 and a section of the surface through P perpendicular to its 
 axis in Q; then ultimately 
 
 Ie ) d 
 sin CPD = 55, =P? or sina. — =p, 
 d t dye 
 or ee geene. J + (| > 
 ee dp 
 
 by integrating which (since =f (p) by the equation to the 
 generating curve), we obtain the equation to the projection. 
 
 Ex. Let the surface be an oblate spheroid ; 
 
 2 pA 
 , a —e 
 to integrate this, make —, P 
 
 8 Rol 1h 
 
 2 
 
 , and the result is 
 
 =U 
 
260 
 
 If the surface be a right cone, the projection of the 
 curve is an equiangular spiral; if a sphere radius a, the 
 equation between the radius vector and perpendicular on 
 the tangent to the projection, is 
 
 1 cosec’a  cot’a 
 
 — —__ a 
 
 10. To determine the shortest line that can be drawn 
 between two given points on a surface of revolution. 
 
 Let OC (fig. 48) be the axis of the surface which take 
 for the axis of z, and a plane AOB perpendicular to it for 
 that of wy; DP a portion of the shortest line=s, wv, y, 3 
 the co-ordinates of P, ON=p, 2AON=9, 
 
 d.s\2  (apy* : 
 pos el EP, a) +f ()}*, 
 if s =f(p) be the equation to the generating curve, and 
 dz ‘ 
 dp =f (p); 
 = Fay VA + ap? + {f' (wv)? = LV, 
 
 ‘ d 
 putting p=a, p=y, <P =p, in order to adapt the expres- 
 P 
 
 sion to the usual formule of the Calculus of Variations. 
 
 dV dV x’ 
 Hence io =10; Pp ee 
 dy 
 
 dp 1/1 + a*p* + Sf" (a) }?’ 
 
 but tS bodys + &c. = 0, for a maximum or minimum; there- 
 
 ev 
 
 fore in this case P = a constant, 
 
 wp ap ds 
 RN erene ae Tomar! Tid Tem oy aft 
 + ap’ + if(a)} p dp 
 
 ds Cc F 1 
 p+ ae = 5 or sin CPDc< ON” as shewn in Prob. 9. 
 
 Hence the shortest line always cuts the meridian at an 
 angle whose sine varies inversely as the distance of the 
 
261 
 
 point of intersection from the axis; and the equation to 
 its projection on a plane perpendicular to the axis of the 
 surface is 
 
 2 am Mel bial 9 aol A 
 gE, 
 
 BOP ENG dp 
 e being the distance from the axis at which the shortest 
 line cuts the meridian perpendicularly. Also for the length 
 ds p dd 
 
 of the shortest line we have — = 
 
 dp c dp 
 
 Ex.1. Let the surface be a right cone, and mz =pV/1 —m’, 
 the equation to the generating line, then 
 
 d Poe ae 
 a £ is p =esecm(p+a), 
 
 the es st to the projection, which is therefore an elliptic 
 spiral. 
 
 Ex. 2. Let the surface be a sphere, and let DP be 
 
 the arc of a great circle, then 
 
 sin CD 
 
 ie 
 sin n CP 
 
 sin D; 
 
 if now the meridian revolve round OC, 
 
 ] 1 
 o —— 
 sin CP ON 
 
 sin Poc 
 
 . 
 > 
 
 hence an arc of a great circle is the shortest line that can 
 connect two points on the surface of a sphere; this result 
 may be obtained from the formula. 
 
 Ex. 3. To determine the shortest line on the surface of 
 an oblate spheroid. 
 
 Let CEB (fig. 49) be the plane of the equator, PE, PN 
 
262 
 
 two meridians, 4B a portion of the shortest line cutting PE 
 perpendicularly at 4. Let the point M be determined by the 
 angle @, so that CT'=p =asin0, ’'M=bcos@; and let A 
 be similarly determined by the angle a, so that its distance 
 from the axis = a sina; then by substitution we get 
 
 ay OND cos’ 9 + 6? sin’ @ 
 = sin@ , 
 
 sin’ @ — sin’ a 
 Now, since a is the least value that 0 can assume, let 
 
 cos 9 = cosa cos Wy, 
 
 i d in 
 Pees ; ee. 4. beet Vals b? + (a? — b°) cos’ a cos” Ws 
 
 ‘ cos a sin Vr ON: cos a sin Wy 
 
 () Dre _Ja cos’ a + b* sin’ a — (a” — b*) cos’ a sin’ Wy 
 
 = aft —c’sin* Wy, 
 
 therefore s = arc of an ellipse, amplitude yy, whose semiaxes 
 
 are a’ =/a? cos?a + b? sin? a, b, and eccentricity =c. At the 
 point B, 6 = 90°, therefore y = 90°, and AB =a quadrant of 
 the ellipse; also 
 
 asina 
 P 
 ZL ABE = 90 - a. 
 
 sin PMA = = sina at the point B, 
 
 For the projection on the plane of the equator, 
 
 dd : ds dp _ sina ds 
 
 since p° — =asina 
 
 do dp’ ms dw ~ asin? @ dy’ 
 
 aD a a V/1—c' sin? 
 adi, asina 1+msin* 
 
 , where 2 = cot*a, 
 
 2 2 
 
 1 a r 1 
 i ere | AE + m sin’ hy) 
 
 ~ asina ENE TE — ¢c’ sin’ PW 
 
‘ p= {( +o) IT, (7, W- SF}, 
 
 employing the notation of elliptic functions (Integ. Cale. 
 p. 208.) 
 
 Cor. If the eccentricity of the generating ellipse be 
 very small, let 
 
 a oe 
 b? 
 
 =e, and cos’acos* Wy =m; 
 
 dp sina \/b? + (a? — b’) cos*acos’*W sina Ry/ Leh me 
 
 diy a 1 — cos’ a cos’ vy l—m a rey 
 sin a 1-m l1—m)(3+m) . 
 7 1) - Ee 4) EMO ACA) wee 
 1—m™m Q 8 
 sIn a sina o6 ¢& COSs’a 
 = 2 ae a (« - a ee cos’ yy — &e.] : 
 1 — cos“ acos Vy 2 4A 8 
 tan sing 38e é . sing ; 
 . p=tan( : 4 —y (6-5 | +o(ys "| sinacosta+ &e. 
 sina 2 4, 16 2 
 
 The indefinite continuation of the curve will manifestly 
 produce an infinite number of spires similar and equal, and 
 contained between two parallels each at a distance = b cosa 
 from the equator. 
 
 11. If a uniform and perfectly flexible thread of given 
 length have its ends fastened in a surface of revolution whose 
 axis 1s vertical, and rest upon that surface supposed perfectly 
 smooth, to find the curve into which it will be formed. 
 
 Taking the axis of the surface for that of z, and a plane 
 perpendicular to it for that of wy, and proceeding as in Prob. 10, 
 we have, # and & denoting constant quantities, 
 
 to (noe) = J, (h +2) Ae + p* a4 + ta =a minimum ; 
 
264 
 
 do dp\** (dz\* 
 o* % h —_ = k 1 2 —— (=) 
 
 Qa ee +e (7) +z) 
 which, since we have x = f(p) by the equation to the gene- 
 rating curve, is the differential equation to the projection of 
 the required curve on a plane perpendicular to the axis of 
 the surface. 
 
 12. The lines of greatest inclination on the surface 
 \/ : . : : 
 Seif 7) are given by the intersection of the surface with 
 @ 
 
 the cylinder x + y* = a’, where a is a parameter. 
 
 13. The tangent line at the point wyz to the curve 
 
 2 2 2 2 v4 
 @ Yy 20 x y & 
 ne ae ee Ore +—-1 
 a? 2 a a  ¢ 
 x-v yy Y-y 2 2-8 
 1S > et Fe Ser ° 
 a 6° a-—@ a c 
 
 PROBLEMS ON Section VI. 
 
 1. Ir a surface of the second order having a center be 
 cut by a plane which always passes through a fixed point, to 
 find the locus of the centers of all the plane sections. 
 
 Let Aa’? + By’ + Cz’ = D be the equation to the surface, 
 z—c=A(w@7—a)+B (y-b) (1) 
 
 the equation to the cutting plane, a, b, c being the co-ordinates 
 of the point through which it always passes; therefore, elimi- 
 nating y, the equation to the projection of the section on xw@ is 
 
 Jez -c—dA'(w-a) + BO? =D. (2) 
 
 z 
 
 . B 
 24 Cx? 
 A x + et 
 
265 
 
 Let h, hk, J be the co-ordinates of the center of the section, 
 then h and 7 are the co-ordinates of the center of the projec- 
 tion; if therefore in equation (2) we substitute # +h for a, 
 and x +/ for x, the projection will be referred to its center 
 as origin, and therefore the coefficients of wv and x’ must 
 each vanish; 
 
 B 
 B”? 
 
 pete i(k 0) AP =1(B bee cy AR = 0, 
 
 B , , 
 Cl + Fe id — 0) + Bb- A (h—a)} =0. 
 
 Multiply the latter of these equations by 4’ and add it 
 
 to the former, 
 
 : : Ah 
 ee Ah+CAl=0, CM pint reat 
 imilar] B re 
 sImMarly, = ie 
 
 Hence substituting these values of 4’ and B’ in (1), and 
 expressing that it is satisfied by h, k, 1, we get 
 
 Ah(h-—a)+ Bk (k- 6b) +Cl(l—-c) =0, 
 
 the required equation, which represents a surface of the second 
 order similar to the proposed one, the co-ordinates of its center 
 being 4a, 3b, de. 
 
 It is manifest that the two surfaces intersect in a plane of 
 which the equation is daw + Bhy+Cez =D. 
 
 2. To find the locus of the middle points of chords of 
 a surface of the second order that has a center, all passing 
 through a given fixed point. 
 
 Take the given point for the origin, and two conjugate 
 diametral planes which pass through it for the planes of z@ 
 and wy, and a plane parallel to the third conjugate plane 
 for that of yx; then the equation to the surface will be 
 of the form 
 
 ax’ + by? +c2x°+2a v+d=0, 
 
266 
 
 Let v= mz, y = nx, be the equations to any chord; then 
 combining these equations with the former, we have 
 
 (am? + bn? + c) 22 + 2a"ms+d=0, 
 
 in which equation the values of = are the co-ordinates of the 
 extremities of the chord; therefore the co-ordinates of its 
 middle point are 
 
 ” 
 am , , , 
 3 vV= MB, Y=H=ns. 
 
 , 
 CS Se Sy ee 
 am? + bn? + ¢ 
 Hence, eliminating m and n, the required equation to 
 the locus is 
 
 an oy by” a ce” + ie 0; 
 
 which represents a surface of the second order, similar to the 
 proposed one, and similarly situated and passing through its 
 center and through the origin. 
 
 8. The curve of contact of a surface of the second 
 order, and of a cone the vertex of which is given, is a plane 
 curve. 
 
 Let aa’ + by +2? 4+2a'a +2by + 2e2+d=0, 
 
 be the equation to the surface which may represent all surfaces 
 of the second order. Then the equation to the tangent plane 
 at a point vys is 
 
 (ax+a)X+ (by+0') V+ (ex4+e) Z+ar+by+cx24d=0. 
 
 Now because the cone and surface touch one another, at 
 the points of contact their tangent planes coincide; and since 
 the tangent plane of a conical surface always passes through 
 its vertex, if h, k, 1 be the co-ordinates of the vertex they 
 must satisfy the above equation ; 
 
 “ (avt+ayjh+(byt+0)k+(exte)l+av+by+e2+d=0, 
 
 which gives the relation among the co-ordinates wv, y, x of the 
 points of contact, and shews that they all lie in the same plane. 
 The curve of contact is therefore a plane curve resulting from 
 
267 
 
 the intersection of the plane whose equation we have just found 
 with the given surface. 
 
 Cor. If the surface have a center, the line joining that 
 point and the vertex of the enveloping cone, passes through 
 the center of the curve of contact. 
 
 Let C (fig. 52) be the center of the surface, 4 the vertex 
 of the enveloping cone, Q2#S the plane of the curve of contact ; 
 and let any plane drawn through 4 and C cut the surface in 
 the line of the second order PQG, and the cone in the straight 
 lines 4Q, AS’; then C is the center of the section PQG, and 
 AQ, AS are tangents to it. Also, by a property of lines of 
 the second order, QS, a line joining the points of contact, 
 is an ordinate to the diameter PG which passes through the 
 intersection of the tangents; therefore QS’ is parallel to the 
 tangent at P, and is bisected in V, and CV.CA=CP’. 
 Since therefore every chord of the curve of contact drawn 
 through the point where a line joining the vertex of the cone 
 and the center of the surface meets the plane of contact, is 
 bisected in it, that point is the center of the curve of contact ; 
 also the plane of contact is parallel to the plane touching the 
 surface at the point where the same line meets the surface, 
 and its distance from C is determined by the equation 
 
 CV.CA = CP’. 
 
 Similarly it may be shewn, that if a cone envelope an 
 elliptic paraboloid, the line drawn through the vertex parallel 
 to the axis of the surface passes through the center of the curve 
 of contact; and the portion of it between that point and the 
 vertex is bisected by the surface. 
 
 4. If the plane of contact of the enveloping cone of 
 a surface of the second order always pass through a fixed point 
 or a fixed line, the locus of the vertex of the cone will be 
 respectively a plane or a straight line. 
 
 Let 2’, y’, x be the co-ordinates of the fixed point through 
 which the plane of contact always passes; then they must 
 satisfy its equation, 
 
 “. (av +a)hs (by +0)k 4+ (c2’4+0)l4+ a a'+b'y'+0'x’+d=0, 
 
268 
 
 which gives the relation among the co-ordinates h, k, 1 of 
 the vertex of the cone, and shews that the locus of the 
 vertex is a plane; and that that plane is the plane of 
 contact with the given surface of a cone whose vertex is 
 in the given fixed point, for the equation may be put 
 under the form 
 
 (ah+a)a'+ (bk +0 )y'+(cl+c)x+adh+bk+cl+d=0. 
 
 Again, let v=mz+ fy y=nz+g, be the equations to 
 the line through which the plane of contact always passes. 
 Then, in order that the plane may contain the line, we 
 must have (Art. 23), 
 
 (ah +a')m+ (bk +b')n + (cl + c’) =0, 
 (ah+a)f+(bk+0)g+ah+0k+cl+d=0, 
 
 which are the equations to the straight line in which the 
 vertex moves, its co-ordinates being h, k, /. 
 
 5. If the plane of contact of the enveloping cone of a 
 surface of the second order always touch a surface of the 
 second order, the locus of the vertex will be another surface 
 of the second order. 
 
 Let the plane of contact always touch a surface of which 
 the equation is 
 
 me+ny +72? 4+2m'e+2n'y+2rx24+5=0......(1)3 
 
 then the equation to a plane touching this surface at a point 
 VY 1s 
 
 (ma+m')a'+(ny+n')y + (retr)s 4+m’'et+n'yt+r's+s=0, 
 
 which must be identical with the equation to the plane of 
 contact, namely, 
 
 (ah +a')a'+ (bk +b)y +(cl tc) +adh+Vk+cl+d=0; 
 
 met+tm ah+a ny+n bk+d 
 
 t 
 
 id eth oleae. eee en | Reet oy ley a. cree. 
 
 metny+re+s ah+bk+cl+d 
 me ees —_________ 
 
 , 
 
 TZ+7 el +c! 
 
269 
 
 and it remains to eliminate v, y, x between these equations 
 and equation (1). The result will be found to be 
 
 m , n’ , ” r , , ‘ 
 {i (ahta)+ (b+ U) 42 (el+e)—Whatkrcl+d| 
 
 m2 mn? “2 1 1 1 
 rar. nea aide on h 4\2 yeni 2 a /\2 
 (= eee 5) > (a +a) +7 (bk +b) +i (else)t, 
 
 which represents a surface of the second order. 
 
 Cor. If the surfaces are concentric and 
 av+by+ex,=1, ma’+ny? +72’ =1, 
 
 their equations, the above result is reduced to 
 2 2 2 
 
 Cie TO ae 
 —h?+—kh?+-—P=1. 
 
 m n r 
 
 6. If two surfaces of the second order have each a prin- 
 cipal section in the same plane, the projection of their curve of 
 intersection on that plane is a curve of the second order. 
 
 Suppose the plane of wy to contain a principal section of 
 each surface, then no odd powers of z can enter into their 
 equations, which will therefore be of the form 
 
 Av’ + By? + Cx? +2Cay + 2A’a+ 2B"y + D=0, 
 aa’ + by? + cz + 2Qcaey +20" a4 +2b' y+d=0, 
 
 and if we multiply these equations respectively by ¢ and C 
 and subtract, we shall find the equation to the projection of 
 the curve of intersection of the surfaces on the plane of vy to 
 be of the second degree in w and y. 
 
 Hence if the axes of two surfaces of revolution intersect, 
 since the plane containing the axes will be a principal plane of 
 each surface, the projection of the curve of intersection on that 
 plane will be a curve of the second order. 
 
 7. To find the diametral surface of 
 Avy + es + yz) = vy, 
 
270 
 
 which passes through the middle points of a system of chords 
 equally inclined to the three co-ordinate axes. 
 
 Here the equation to any chord, if aB-y be its middle 
 point and m4/3 = 1, will be 
 
 w-a=y-B=2s-y=mr, 
 - a}(at+mr)(B+mr)+(atmr)(y+mr)+(B+mr)\(y+mr)} 
 =(a+mr) (B+ mr) (y +mr), 
 or m7 +(at+B+y—3a)m' r+ faBtay+By—-2a(atB+y) mr. 
 +{aBy -a(aB+ay+By)$=0, 
 or Ar? + Br? +Cr+D=0; and if p, —p, q be the roots, 
 —Ag=B, —-Ap=C, Apq=D; «. AD=BC; 
 . aBy —a(aBt+ay + By) 
 =(a+B+y —-3a)$aB+ayt+ By -2a(at+ B+y)} 
 
 the required equation. 
 
 8. The sum of the squares of the projections of an 
 ellipsoid on any three planes at right angles to one another, 
 is equal to the sum of the squares of its projections on its 
 principal planes. 
 
 Let the equation to the surface referred to any three 
 rectangular axes through its center be 
 
 av’ + by +c22 + 2ay%4+20ax+2cuy=d, (1). 
 Since at those points in the surface for which 
 cz +ay+ba=0, (2) 
 dz dz 
 dav dy 
 are parallel to the axis of x; hence, eliminating = between 
 
 (1) and (2), the equation to the projection of the ellipsoid 
 on the plane of ay is 
 
 become zero, the tangent planes at those points 
 
 a (ac — b°) + y? (be — a”) + 2ay (ec —a'b’) = de; 
 
271 
 
 .. square of area of this projection 
 a’ dc” nw’ d’c 
 
 ~ (ac — b”) (be — a”) - (ec’- ab’)? = re 
 
 (S,5 S25 83, being the roots of the cubic in Art. 146) therefore 
 sum of squares of projections on the co-ordinate planes 
 
 w (a’ B? af ary? + By’) 
 
 where a, 3, y are the three semi-axes of the ellipsoid. 
 
 wd’ (a+b+ec) 
 a 8 82 83 * 
 
 9. If O be a given point in a surface of the second 
 order and OA, OB, OC any three chords passing through O, 
 mutually at right angles, the plane ABC will always pass 
 through a fixed point in the normal at O. 
 
 Let the equation to the surface be 
 au’ + by? + cz? + cay +c'2s=0, 
 
 the co-ordinates being rectangular, and the axis of z a normal 
 at the origin. Let the surface be referred to three new rect- 
 angular axes ; 
 2. a(ma’ +ny +r2')+ b(m'a' +n'y' +17'')?+ c(m' a’ +n"y' +7"s")? 
 +0 (ma! +ny' +73) (ma +n'y' +9'x') +0" (ma +n’y' +72’) =0. 
 
 Let h, k, 1 be the parts of the axes of a’, y’, x’ intercepted 
 by the surface 
 
 M9 oe c'm m’) a em” = 0, 
 
 “. h (am? + bm? + em 
 k (an? + bn? + cn”? + cnn’) + cn" = 0, 
 Lar + br? + er? err’) +67 = 03 
 therefore the equation to a plane passing through the ex- 
 tremities of h, k, 7 is 
 Ri wiloistetnk 
 and referred to the original axes, it is 
 metmy+m’s netnytn’s retrytr’s 
 rey | 1g Ie i 
 
 Il 
 pd 
 i) 
 
272 
 
 and putting «=0, y =0, to find when it meets the axis x, 
 we have 
 
 ” A Ad 
 
 \=1 (a+b+c) +c’ =0 
 — + —+—]j =1, or x(a c= 0, 
 (s+ E45 ’ C) + 
 
 consequently the plane cuts the normal in a fixed point. 
 
 10. An infinite number of surfaces of the second order 
 may be made to pass through the sides of a given quadri- 
 lateral in space. 
 
 Let ¢=0, w=0, v=0, w=0, be the equations to four 
 planes OAB, BAC, CAO, OBC (fig. 11) forming a tetrahe- 
 dron; then the equations to the four edges AB, BC, CO, OA 
 
 forming a quadrilateral in space, will be 
 
 t=0 u=0 w=0 v=0 (1) 
 =01 > ep = Ole tell Pele 
 
 If therefore the equation to a surface of the second order be 
 tw =p.Wv, it is evident that every one of the four straight 
 lines (1) will be situated in the surface throughout its entire 
 length ; and as uw is arbitrary, an infinite number of surfaces 
 may so pass through the same four lines. 
 
 11. <Any tangent plane to a two-sheeted hyperboloid will 
 cut off from the asymptotic cone, a cone of constant volume ; 
 and the point of contact with the hyperboloid will be the 
 center of the plane section of the asymptotic cone. 
 
 12. If through any point of a hyperboloid of one sheet 
 the two generating lines that intersect in that point, and a 
 diameter of the surface, be drawn to meet any tangent plane 
 of the asymptotic cone, a pyramid will be formed of constant 
 volume. 
 
 13. The locus of the centers of all sections of the surface 
 Ax’? + By? + Cz’ =1 that are at the same distance d from 
 the center has for equation 
 
 (Aa? + By? + CC) = (Aa? + By? + C? x”) &. 
 
 14, The area of a section of an ellipsoid made by a 
 
273 
 
 plane through its center such that a perpendicular p to it 
 makes angles a, (3, yy with the axes of the ellipsoid, 
 
 mabe awabe 
 Pp \/ a2costa + b’cos® (3 + c°cos*ry 
 
 15. The locus of the intersection of tangent planes to 
 an ellipsoid drawn at the extremities of a system of conjugate 
 diameters, is an ellipsoid concentric with the original one, and 
 
 having its axes greater in the ratio of / 3 to 1. 
 
 ProBLEMs ON Section VII. 
 
 1. ‘To find the equation to the cylindrical surface which 
 envelopes a given oblate spheroid. 
 
 Let the center of the spheroid be the origin, and the 
 plane of sw that which contains the axis of the spheroid, 
 and a line through the origin to which the generating line 
 of the cylinder is always parallel; therefore the generating 
 line in any position will be w=mz+a, y=. 
 
 Let c? (a + y’) + a’2” =a’c’ be the equation to the 
 : dz dz 
 spheroid, therefore the equation m——+n-——=1 becomes, 
 dx dy 
 since n = 0, 
 mca + a?x =0, which together with c? (a + y’) + a’s” =a’c* 
 
 are the equations to the directrix; and if between these and 
 the equations 7 = mx +a, y = 3, we eliminate x, y, x we find 
 
 3? (a2 + mc’) + aa? = a? (a? + mc’). 
 
 Therefore, restoring the values of a and 3, the equation 
 to the surface is 
 
 y? (a + m?c’) + a? (@ — mz) = a’ (a? + mc’). 
 
 If we make x=-—c, we shall obtain the equation to 
 18 
 
O74 
 
 the curve in which the cylinder meets the horizontal plane 
 on which the spheroid stands; ~ this gives 
 
 yy («+mc) 
 
 @ atmo” 
 which is the equation to an ellipse, the origin, that is, the 
 point on which the spheroid stands, being the focus. Hence, 
 an oblate spheroid placed in the Sun on a horizontal plane 
 
 stands in the focus of its shadow. 
 
 2. If a cylindrical surface be circumscribed about a 
 surface of the second order of which the equation is 
 Aw’ + By’ + Cz’? =1, and if R be the length of the semi- 
 diameter to which the cylinder is parallel, and A, uw, v the 
 angles which it makes with the axes, the equation to the 
 cylindrical surface is 
 
 Av’ + By +Cz? -1= RF (Axcosdy\ + Bycosp + Cz cos pv)’. 
 
 3. To find the equation to the conical surface which 
 envelopes a given surface of the second order. 
 
 Let the given surface be Av? + By’? + Cz’? = D, then the 
 equations to any generating line through the vertex (a, b, c) 
 will be 
 
 and to find the intersection of this with the surface we have 
 A(a+ilry?+B(b+mr)’?+C(e+nr)?-D=0, 
 
 and since the values of r must be equal, we have between 
 l, m and nm the relation 
 
 : (Aal+ Bbhm + Cen)? 
 = (AP + Bm’ + Cn’) (Aa? + BB? + Ce* — D), 
 
 therefore substituting for 1, m, m their values from (1), 7 dis- 
 appears and we get the equation to the enveloping cone 
 
 {Aa (wa) + Bb(y - b) + Ce(z -c)}? 
 = {A (w-a)? + B(y—b)* + C (w—c)*t (Aa? + BL?+ Cc? D) 
 
O15 
 
 which reduces at once (agreeably to Cor. Art. 165) to 
 (Aa? + By’ + Cs -— D) (Ad? + Bb? + Ce? - D) 
 = (dav + Bhy + Cex — D)’. 
 
 It is evident that in the same way the enveloping cone to 
 any surface represented by an algebraic equation F(a, y, x) =0 
 may be determined, by finding the equation of condition between 
 l, m, and m that F(a +/lr, b+mr, c+ nr) =0 may have a 
 pair of equal roots, and then eliminating /, m, and n by means 
 of the equations (1). 
 
 The cone touches the surface along the curve in which it is 
 intersected by the plane represented by Aaw + Bhy+Cezx=D 
 obtained by substituting for p and qins—c=p(v-—a)+q(y—8). 
 Also if S'= AC (fig. 52) the length of the line joining the 
 vertex of the cone and the origin, and R = CP the portion 
 of that line which forms a semidiameter of the surface, it 
 may be shewn that 
 
 Aa tb be + -Cu = DP. 
 
 which may be substituted in the above equation to bs 
 conical surface. 
 
 Cor. If instead of referring the surface to its axes, 
 we refer it to a system of conjugate diameters, one of which 
 passes through the given vertex and is taken for the axis 
 of z, the equation to the plane of contact will be Ces -D=0; 
 and the equation to the enveloping cone will be found to be 
 
 ree, t 
 (x — c)? = (5 ~ Fa (Aw? + By’). 
 
 4. If a cone envelope an elliptic paraboloid, the line 
 joining its vertex and the center of the curve of contact is 
 parallel to the axis of the paraboloid, and bisected by the 
 surface; and the volume of the cone is “rds of the volume 
 
 3 
 of the enveloped segment of the paraboloid. 
 
 5. If lines be drawn from the center of an ellipsoid 
 parallel to the generating lines of an enveloping cone, a 
 
 18—2 
 
onG 
 
 conical surface will be formed intersecting the ellipsoid in two 
 planes parallel to the plane of contact. 
 
 6. One of the most useful applications of the general 
 method of determining conical surfaces, is to find the stereo- 
 graphic projection of a curve upon any plane; that is, the 
 trace upon the plane, of a conical surface, called the optical 
 cone, whose vertex is the place of the eye, and directrix the 
 given curve. 
 
 If we take a line drawn through the place of the eye, 
 perpendicular to the plane of projection for the axis of 2, 
 the equation to the optical cone will be 
 
 feeb ash 
 F+C +C 
 
 which results from the general equation to conical surfaces, 
 by making a= b= 0, and changing the sign of c, that is, 
 supposing the vertex to be situated below the plane of vy, 
 If in this equation we make sz constant, we obtain the 
 equation to the stereographic projection of the given curve 
 upon any plane perpendicular to the axis of =; and if in 
 the result we make c infinite (so that the projecting lines 
 may all become parallel to one another and _ perpendicular 
 to the plane of projection), we shall find the equation to 
 the orthographic projection, which is the sort of projection 
 so frequently employed in every part of this subject. 
 
 When the curves to be projected are situated on the sur- 
 face of a sphere, the term stereographic is usually con- 
 fined to the case where the eye is situated in the pole of 
 the great circle whose plane is the plane of projection; 
 and when the eye is in the center, and the plane of pro- 
 jection is a tangent plane to the surface, the projection is 
 called gnomonic. We shall now illustrate the method by 
 the following examples. 
 
 7. To find the stereographic, gnomonic, and orthographic 
 projections of any circle of a sphere. 
 
 Let the center of the sphere be the origin, and a plane 
 
277 
 
 perpendicular to the plane of the circle to be projected, 
 the plane of zw; then 
 
 P+y+e =r, s=axrt+d, 
 are the equations to a circle of the sphere, forming the 
 directrix of the optical cone ; 
 also w=a(z+c), y=B(e+0), 
 
 are the equations to the generating line; hence, eliminating 
 Xv, Y, x, between these four equations, we find 
 
 acat+b bh+e b+e 
 oe! P2e yerrmees $A ah ef! ) a : 
 
 5) aT 1—da l-—adaa 
 
 “. (b +c) (a’ + PB’) + (aca + 6)? =7* (1 - aa)’; 
 or, restoring the values of a and #3, 
 (b+ c) (a? +y)t+laca+b(z+o)P=r(z+e-az)’, 
 
 the equation to the optical cone, from which those to the 
 projections may be deduced, as follows. 
 
 In stereographic projection, the plane of projection passes 
 through the center, and the eye is situated in the surface of 
 the sphere, therefore s = 0, c =7; 
 
 “ (b+r) (a +y) +7 (aa +b) =r (* - aay’, 
 
 2ar 2 Pp — OD 
 vC=r 
 
 y +b Pee Oe 
 
 or w+ + 
 
 ° ° ° Ue Te er ae ee 
 the equation to a circle, whose radius = oa Vr (1 +a’) — 0, 
 Pak 
 
 ar 
 
 r+b 
 
 and distance of its center from that of the sphere = 
 
 In gnomonic projection, the plane of projection touches 
 the surface, and the eye is situated in the center of the sphere, 
 therefore s = 7, c=0; 
 
 Rw +y)+Prar (r-azy’, 
 
 or a (0 — ar’) + By’? = 7 (7 — 0’) - 2rav; 
 
278 
 
 the equation to an ellipse, hyperbola, or parabola, according 
 as b>, <, or = a’7". 
 
 In orthographic projection c= ©, # =0; 
 . vw +y + (ax + b) =2°, the equation to an ellipse. 
 
 It is to be observed that the optical cone, since it meets 
 the surface of the sphere in a plane curve, will meet that 
 surface again in a plane curve (Art. 78); so that the stereo- 
 graphic projection relative to any point, of any circle of a sphere 
 upon the surface of the sphere, is a circle. 
 
 In the same manner it may be shewn that the stereo- 
 graphic projection of any plane section of a paraboloid of 
 revolution upon a plane perpendicular to its axis, isa circle, 
 the eye being placed in the vertex; or that the stereographic 
 projection of any section of an oblate spheroid upon the plane 
 of the equator, is a circle, the eye being placed in the pole. 
 We may employ the same method also to compare any angle 
 on the surface of a sphere with its different projections ; this, 
 however, may be done more simply by geometrical con- 
 siderations, as will be seen in the following problem. 
 
 8. The gnomonic, orthographic, and stereographic pro- 
 jections of the rhumb-line, which is a curve of double curvature 
 traced on the surface of a sphere so as to cut all meridians 
 under the same angle, are three of Cotes’s spirals. 
 
 First to find the gnomonic projection. 
 
 Let y Ba, (fig. 55) be the plane of projection perpendicular 
 to BC, T¢ its intersection with a plane touching the sphere at 
 a point P in the plane of za, then T¢ is parallel to By; Pt 
 a tangent to the rhumb-line at P, Q the projection of P de- 
 termined by producing CP to meet the plane wBy. Then 
 ZTQ¢t is the projection of T'Pt; 
 
 A ES Me 
 
 CB 
 and tan 7'Q TQ TQ TP TQ tan 7'Pt CQ tana, 
 
 if a =the constant angle at which the rhumb-line cuts the 
 
279 
 meridian. Let BQ=p, p=perpendicular from B on Qt 
 which touches the projection of the rhumb-line at Q, BC =a, 
 
 ei p atana Li cosec? q/\* cot’a 
 
 Jip V/p+a ep p” rs ate? 
 
 the equation to one of Cotes's spirals. 
 
 Secondly to find the orthographic projection. 
 
 Let # Cy be the plane of projection (fig. 56), T?¢ its inter- 
 section with a plane touching the sphere at a point P, situated 
 in the plane of za, then 7'¢ is parallel to Cy; Pt a tangent 
 to the rhumb-line at P, N the projection of P, thenz TN¢ 
 is the projection of 7'P¢; 
 
 Ls Gene 0) Ege, fate iy GP 
 
 t = SS ee ee SC >| . f = 3 
 and tan 7'N¢ TN Ty’ TP pn’ T'Pt py ane 
 or if CN = p, p= perpendicular from C on N# which touches 
 the projection at JV, 
 
 p atana Dee cosec’ ai cot? a 
 
 ye RU ar rR ca 
 the equation to the hyperbolic spiral, as already found p. 260. 
 
 Thirdly to find the stereographic projection, 
 
 Let CO = CP be perpendicular to the plane of projection 
 «Cy (fig. 56) ; join PO meeting the plane of projection in Q, 
 which is therefore the projection of P, and 2 7'Qt of T'Pt. 
 Then Z TPQ =complement of CPO =complement of COP 
 =PQT, .«. PT=QT;; and since 7J'¢ is perpendicular both to 
 PT and QT, .. 2TQt=T7Pt=a; therefore since the angle 
 between the tangent and radius vector is constant, the curve is 
 an equiangular spiral, its equation being p = p sin a. 
 
 Gor. It appears, that if an angle a on the surface of a 
 sphere have one of the arcs which contain it, a meridian, and if 
 ry denote its gnomonic, w its orthographic projection, and @ the 
 are of the containing meridian intercepted between its vertex 
 and a plane through the center parallel to the plane of projec- 
 
280 
 
 tion, then tan y=sin@tana, tanw= tana; also tana 
 
 sin 
 = /tanw. tan ry, that is, the tangent of the stereographic 
 projection is a mean proportional between the tangents of the 
 orthographic and gnomonic projections of the same angle. 
 
 9. To find the equation to the conoidal surface gene- 
 rated by a horizontal straight line which constantly passes 
 through a Helix, and the axis of the vertical cylinder on 
 which it is traced. 
 
 The equations to the Helix are (Prob. 2, p. 254) 
 w 
 e+y=a’, z=nacos'—; 
 a 
 
 let the equations to the generating line be s=, y= aa, 
 then eliminating a, y, 2, a +a°’2 =a’, 
 
 & ] 1 
 
 Te ee ig V1 a 
 therefore, restoring the values of a and £, the equation to the 
 surface is 
 
 1 
 
 ie cose =natan—'aq; 
 
 Z= na tan) ; 
 
 c 
 
 This is the surface presented by the inferior superficies of a 
 staircase, attached to a vertical column round which it winds. 
 
 10. To find the equation to the surface generated by 
 the revolution of any straight line about a fixed axis. 
 
 Let the fixed axis be taken for the axis of zx, and let the 
 equations to the directing straight line be 
 
 vw=Azth, y= Brr+k, 
 and those to the generating circle x = 2, a + y° =a sethen 
 eliminating a, y, x, we get (A468 +h)’?+(BB+k) =a, or, 
 restoring the values of a and #, | 
 (Az +h)? +(Bs+kye=a'’s+y’, 
 or vw +4? — (47 + B’) 2 -2(AR+ Bh) 2 =h' + k’, 
 
281. 
 
 which is the equation to the surface, and manifestly represents 
 a hyperboloid of one sheet with its center in the axis of gz. 
 If in this equation we make w = 0, or y = 0, in order to find 
 the equation to a section through the axis, the resulting 
 equation represents a hyperbola; therefore the surface is a 
 hyperboloid of revolution. Also if, retaining one of the above 
 equations to the directrix, we change the signs of the second 
 member in the other, the equation to the surface is not altered; 
 therefore the same surface may be generated by two different 
 straight lines, and through any one of its points two straight 
 lines may be drawn so as to entirely coincide with it. This 
 agrees with Art. 73. 
 
 If we take for axis of x the shortest distance of the 
 directrix from the fixed axis, the equations to the directrix, 
 since it is parallel to the plane of yz, will be r=h, y= Bx; 
 therefore putting 4=0, k=0, the equation to the surface 
 (reckoned from its center as origin) becomes 
 
 e+ yy? — Bs? = h’. 
 11. An oblate spheroid revolves about any diameter, to 
 
 find the equation to the surface which envelopes it in every 
 position. 
 
 Let w=mz, y= nz be the equations to the diameter, 
 c (a + y’) + ax? =a’ the equation to the spheroid ; 
 then the equation 
 (y—nz)p—(# —mz)q+my —nx=0 
 
 becomes (my — nx) (a? — c?) =0; so that the equations to 
 the directrix are 
 
 C(#+7)+asx =a, nx=my, 
 and the equations to the generating circle, 
 s+me+ny=2, wt+y +2 =a; 
 
 eliminating w, y, x between these four equations, we find 
 
 a? (a ~ c°) (m? + n°) = (Bae Va =a)" 
 
282 
 
 therefore, restoring the values of a@ and #, the required 
 equation is 
 a? (a + y® + 3? — 0?) (m? + 2’) 
 
 =S[(st+me+ny)/ae-@ —cV/a —a —y — #°) 
 
 12. The locus of the normal to the surface y = # tannz 
 along a generating line is a hyperbolic paraboloid. 
 
 PROBLEMS ON SeEctTion VIII. 
 
 Tue following are some examples of finding the equations 
 to twisted and developable surfaces, and envelopes. 
 
 1. To find the equation to the twisted surface of which 
 the directrices are two vertical circles having the opposite sides 
 of a horizontal parallelogram for diameters, and a straight line 
 passing through the center of the parallelogram perpendicular 
 to the planes of the circles. 
 
 Let the center of the parallelogram be taken for origin and 
 its plane for that of wy, and the rectilinear directrix for the 
 axis of y; then the equations to the three directrices will be 
 
 c=0 2=0, 
 y=-b (#-a)+xX?=7, 
 y=4+b (#+a)?+2? =2". 
 
 Also let the equations to the moveable line be 
 
 w=a(y-B), x=y(y- 8), 
 
 so that it already fulfils the condition of meeting the axis of y, 
 and one of the parameters consequently is eliminated; then, 
 expressing that it passes through each of the circles, we have 
 
 CC eae ane (1); 
 12 (6 B) 42h" 4 oO = Be ey Cees 
 
283 
 
 therefore, by subtraction, 
 B (ba? + aa + by’) =0; 
 
 and rejecting the value 8 = 0 which would make the moveable 
 
 line always pass through the origin and so generate an oblique 
 cone, we have 
 
 a” + 9° +o = Qe» (2)s 
 
 by virtue of which, either of the equations (1) gives 
 | aa (b®? — B’) = b(7* — a’)...(3); 
 
 and it remains to eliminate a, 9, yy between (2) and (3) and 
 the equations to the generating line. Substituting the values 
 of a and vy given by the latter in (2) we find 
 
 b a? + 2° a w 
 
 =yt+- and then a = —— ———.; 
 PAL, Gia alas b wv + 3? 
 
 therefore, substituting these values of a and @ in (3) and re- 
 ducing, we find for the equation to the surface 
 
 faany + b (a + 2°) }? = br? ax + b? 2? (7? — a’). 
 2. To find the equation to the developable surface gene- 
 rated by a straight line which constantly touches a Helix. 
 
 The equations to the Helix being 
 x ns 
 v=acos—, y=asin—, 
 na na 
 
 at a point for which x = a the equations to its tangent are 
 
 a Late ne 
 v — a cos—— = — — sin — (z — a), 
 na n na 
 
 hag Oi 1 a 
 y — asin — = — cos — (#—- a); 
 Nad nm na 
 
 a i oeet a ar\/ a + 2_ ge — aw 
 , @ CoS — + Y sin — =a, and tan — Eyed da Yn & Gs 
 na na 
 
 na y - a 
 
284 
 also, adding the squares of the above equations, we find 
 1 
 2 2 2 \2 
 ¢ + = ys eS, : 
 vty —a 5 (% -a) 
 
 therefore the equation to the surface is 
 
 ——_——,, a/xv+y—-a—2x 
 n/ x +y°—-a@’=2%—-—natan"' (oy eae : 
 y—a 
 3. To find the nature of the curve traced upon a cylin- 
 drical surface, when the surface is made plane. 
 
 Let aw, y, = be the co-ordinates of any point in a curve 
 traced upon a cylindrical surface; then the equations to the 
 generating line which passes through that point are 
 
 zg —-zs=m(a'—-2), y—-y=n(a'—-2@); 
 
 and the equations to the tangent line of the curve are 
 ped pike its Maenacly Be 
 Pitts pe CED yy =— (a -2); 
 
 let 7 denote the angle between these two lines, y the in- 
 clination of the generating line to the axis of w, and s the 
 length of the curve; then (Art. 30) 
 
 cost = cosy ( La +1) 
 i= cosy (m— +n — tT. 
 TO da 
 
 Now when the cylinder is developed, the generating lines 
 preserve their parallelism, and may therefore be taken for 
 the ordinates ; let a line perpendicular to them be the axis 
 of the abscissee, and a, (3, the co-ordinates of the point in 
 question ; 
 
 d d d 
 then “P = cos , bes oP — cosy (m S24 nts 1); also 
 
 1+ 68+ (= 65) +8) ~ Ge +2) 
 
285 
 
 : ds\* : 
 each being the value of (=) ; and if between these and 
 
 wv 
 
 the two equations to the curve we eliminate a, there will 
 remain a differential equation between a and 3 which will 
 be that to the required plane curve. But if the latter 
 equation be given, that is, 8 = f(a), we may eliminate « 
 between the two above equations, and there will remain a 
 differential equation between a, y, and zs, which, with the 
 given equation to the surface, will be the equations to the 
 required curve of double curvature. 
 
 Cor. If the curve on a plane be a straight line, the 
 angle 7 is constant ; 
 
 Pets (m dz dy : "ds 
 eee CO == —— Se 
 Y an 2) en } COS 2 
 
 ve 
 
 is the equation for determining its nature, when the plane 
 is applied to a cylinder ; ae curve is called the Helix, 
 and we have for its length, by integration, 
 
 cos y (ms + ny + #) = cosi(s + C). 
 It is manifestly the shortest line which can join two points 
 on the surface of the cylinder. 
 
 4. To find the nature of the curve traced upon a 
 conical surface, when the surface is made plane. 
 
 Let a, b, c be the co-ordinates of the vertex of a conical 
 surface, v, y, x co-ordinates of a point in a curve traced upon 
 it, and 7 the distance of that point from the vertex. When 
 the cone is developed, the generating lines become radii 
 vectores, and if @ be the angle which x forms with some fixed 
 
 : ds\’ 
 radius vector, equating the two values of (==) » we have 
 a 
 
 ESE ACI ale ee MOEN Nee LAE 
 ae (a) + (3) fo uo a) +13 (5) : 
 gun niente) (yi 2b)" 6 (iC) 
 
 from which, together with the two equations to the curve 
 we may eliminate «, y, x, and there will remain a relation 
 
286 
 
 between x and 0, the equation to the plane curve when the 
 surface is developed; or, if » = (0), and the equation to the 
 conical surface be given, we may eliminate r and 0, and so 
 arrive at the remaining equation to the curve of double 
 curvature resulting from the application of a plane on which 
 a given curve is traced to a given conical surface. 
 
 Cor. All curves which make a constant angle with the 
 generating line of the cone are characterized by the equation 
 
 dr 
 
 Ber ak i, and become equiangular spirals when the surface 
 s 
 
 is developed. Also all curves which become straight lines 
 when the surface is developed, and which therefore are the 
 shortest lines which can connect two points on its surface, are 
 
 d 7 A gps Sak 
 characterized by the equation aes /r — a’, a being the 
 de dex 
 
 length of the perpendicular let fall upon the line from the 
 vertex as origin, or 
 
 1 ee Bag SE a é 2 2 (Jilin yoda 
 zs (5=) ey (54) aaV? See et 
 or finally, by developing this equation, 
 
 Ot) eed FOS ee 3 Otter 0 \. es dz\* (dy\? 
 Ca ea) + (ye 9s) = & {1+(3) + (5) . 
 
 When the surface is a right cone on a circular base, the 
 problem admits of a very simple solution, as will be seen 
 in the following instances. 
 
 5. A given curve being traced on the surface of a right 
 cone whose base is a circle, to determine its equation when the 
 surface is developed. 
 
 Let P (fig. 58) be a point of the curve situated on the 
 generating line CR, Q its projection on a plane perpendicular 
 to the axis; then our object is to find a relation between 
 CP =r, and the angle which CP makes with CA, when the 
 cone is developed ; call this angle 0, and let 
 
 CQ =7', QCx = 6’, m=cosec ACD; 
 
287 
 
 now, when the cone is developed, @ is subtended by a circular 
 are, length AR; 
 
 ~.0.4C =0'.AD, or, 0 =m@; 
 also CP.sin PCD = CQ, or, est 
 
 if, therefore, f(7’, 0°) = 0 be the given equation to the locus 
 of Q, F(=, m8) =0 is the equation to the curve when 
 
 the surface of the cone is developed. 
 
 Conversely, if f(r, 9) =0 be the equation to a curve traced 
 in a circular sector, and the sector be formed into a right cone, 
 the equation to the projection of the curve on a plane perpen- 
 
 , 
 
 dicular to the axis of the cone will be f( mr", =) = 0. 
 m 
 
 Ex. 1. <A right cone being intersected by a sphere whose 
 center is in its surface, to determine the equation to the curve 
 of intersection when the cone is developed. 
 
 Let B in the plane of xa be the center of the sphere (fig. 58) 
 BC =c, a = its radius, a = 4 vertical 4 of the cone, then 
 
 (c sna — w)*?+ (ccosa—32)?+ y= a 
 is the equation to the sphere, and 2° tan?a = 2 +y’ is the 
 
 equation to the cone; therefore, the equation to the pro- 
 jection of their intersection on the plane of wy is 
 
 : 
 c+ (a + y’)§1 + cot? a} -2ca sina — 2¢ ts Sey a+ y? =a’, 
 sina 
 and its polar equation is, making w = 7" cos 6’, y = 7’ sin 6’, 
 c? + 7” cosec’a — 2c7'sin a cos 0’ — 2cr’ cosec a cos’?a = a. 
 But m = coseca, r=? coseca, 0 =m@; 
 
 therefore, when the cone is developed, the equation to the 
 curve of intersection is 
 
 yr? — 2cr fsin? a cos m@ + cos? a} +c-—a@’=0. 
 
 Ex. 2. If a semicircle be described upon the bounding 
 
288 
 
 radius of a quadrant, to find the equation, to its projection 
 upon a plane perpendicular to the axis, when the quadrant 
 is formed into a cone. 
 Tv ° ay" 
 Here m= 27 => ae 4; also the equation to the semicircle 
 is r = a cos 0, therefore the equation to its projection is 
 , Sl 1 1 , 
 r=7acos;6. 
 
 6. To determine the shortest line on the surface of an 
 
 ellipsoid. 
 
 av 7 
 Let — + a + — =1 be the equation to the surface; then 
 a 
 1H q e e 
 p=- bia a q= = es and the equations to the shortest line 
 as & 
 
 Ly eae ving ad’ x ; 
 F papel waka y By, 7 =<" where a” = a &c., the arc s being 
 Bs s 
 
 an independent variable; also we have 3’ sol +qy’ which gives 
 
 wx yy | 88 dle ne 
 
 —— hk a = 01), ands a 
 
 v wo y y"" 2 3"! y pe ar y 2 ) 
 ag EY ate pee SE ae Tee | ey 
 a B Y x z 3° f @) 
 
 rom yw ru ” 
 
 REAPER M EA od ig Fc 
 
 a p + Po Be *y” 
 
 Hence multiplying this by equation (2) so as to eliminate 
 ys" 
 x 
 
 and integrating, we get 
 
 Be A Bee ois ae : 
 Ce eeeaiad 
 oP) NAG: iG Cc 
 
 between which and equation (1) together with the equation 
 to the surface if we eliminate one of the variables, we shall 
 obtain the required differential equation of the first order 
 to the shortest line, or geodesic line as it is called. 
 
 ? 
 
289 
 
 7. The direction of the shortest line at any point of an 
 ellipsoid is determined by the condition that the product 
 - of the semidiameter of the surface which is parallel to that 
 direction, multiplied by the perpendicular from the center on 
 the tangent plane at the same point, is invariable. 
 
 For if a semidiameter D be drawn parallel to the tangent of 
 the shortest line at vyz, wv’, y’, x’ will be the cosines of the angles 
 which it forms with the axes of the ellipsoid, and we shall have 
 
 Sy ad | eg a sean 
 73 +—_= alii chats Coolie aii i 
 from Prob, 3, p. 234, P being the perpendicular from the 
 center on the tangent plane at wyz; therefore the result of 
 the preceding Problem becomes 
 
 PxDe=C. 
 
 8. The following are examples of Envelopes, when the 
 equation to the series of surfaces contains only one parameter. 
 
 Ex. 1. To find the equation to the envelope of all right 
 cones of a constant volume, whose axes are in the same 
 straight line, and bases in the same plane. 
 eee 
 
 b 
 
 The equation to any cone is =@ , where 
 
 a = altitude, 6 = radius of base, and the origin is in the center 
 of the base. Let its volume always equal that of a hemi- 
 sphere, diameter 3c; 
 
 wah’ 7 (8e)? PM aon 27 ¢% J a+ y? 
 %3 : Saeco or ety eae | Le (i-Y=*"), 
 differentiate with respect to the parameter 6, 
 3 3 
 et — — + ee Ja +y, or b == Val ays 
 a a (1 =) = a , the equation required. 
 
 = 2b 276 2) 9c 
 ADV ae carr, Cr abr 
 19 
 
290 
 
 are the equations to the characteristic, that is, determine the 
 radius and position of the center of the circle, in which 
 the cone, the radius of whose base = b, is intersected by the 
 consecutive cone of the same volume. 
 
 Ex. 2. Ifa series of planes, passing through a fixed point 
 in the axis of x, have their traces on the plane of wy all of the 
 same length ; to find the equation to the developable surface 
 formed by their intersections. 
 
 If c= distance of the fixed point from the origin, a = length 
 of the trace on the plane of wy, and a= the angle which it 
 makes with the axis of x, the equation to the plane is 
 
 v 
 bask 
 acosa asina ¢ 
 
 which being differentiated with respect to a gives wtan’a = y; 
 therefore substituting in the equation to the plane, we get 
 
 Ps 1 
 — + — (#3 + y3)§ = 1, 
 c hua 
 
 the required equation to the surface. 
 
 The equations to the characteristic in this case are 
 
 x y 
 -sePa+-=1, == tan’®a. 
 a c av 
 
 Ex. 3, If the center of a sphere whose radius =a, 
 describe a curve in the plane of «xy, to find the equation 
 to the annular surface which envelopes the sphere in every 
 position. 
 
 Let a and £3 be the co-ordinates of the center of the 
 sphere in any position, and y=f(a#) the equation to the 
 curve which it describes, therefore (3 = f(a) ; 
 
 . (w-a)’+ fy -fla)P+=a 
 is the equation to the surface, and by differentiating with 
 respect to a, 
 
 w-at fy—f(a)}f’ (a) =03 
 
 these are the two equations to the characteristic, whose 
 
291 
 
 position and magnitude depend upon a; and if we elimi- 
 nate a between them, we find the equation to the envelope. 
 The characteristic in this case is manifestly a circle of con- 
 stant radius whose center is a point in the curve which 
 forms the axis of the surface, and whose plane is perpen- 
 dicular to the tangent line at that point, as is expressed 
 by the above equations; hence its nature is entirely inde- 
 pendent of the curve whose equation is y= f(x); it is also 
 the curve of greatest inclination, for its tangent line at every 
 point is perpendicular to the intersection, with the plane 
 of wy, of the plane touching the surface at the same point, 
 
 therefore its equation is (p. 257) are The remaining 
 equation is that to the envelope since the characteristic is 
 situated upon it, which we obtain by observing that the 
 normal always meets the plane of zy in the curve which forms 
 the axis of the surface, and the length is equal to the radius 
 of the generating sphere ; 
 
 .=eVYltps+g=a, (Art.110) or #1 4+p'+q) = a3 
 
 this might also have been obtained by eliminating the 
 arbitrary function from the two equations to the characteristic ; 
 hence the integral of the equation z°(1 + p® + q*) =a’ is re- 
 presented by the system of equations to the characteristic. 
 
 Ex. 4. If the vertex of a right cone describe a given 
 curve in the plane of wy, and its axis be always perpen- 
 dicular to that plane, to find the equation to the surface 
 which touches and envelopes it in every position, 
 
 Let a and £ be the co-ordinates of the vertex, y = f(«) the 
 equation to the curve which it describes, therefore 3 = f(a) ; 
 a = tangent of the angle which the side of the cone makes 
 with the plane of wy; then the equation to the surface is 
 
 Pv 
 (@- a)? + fy -f@}= 55 
 differentiate twice successively with respect to a, 
 
 . Sy-fla)t f'(a)+a—-a=0, fy-f(a) $f" (a)-1- ff (a) }?=05 
 19—2 
 
292 
 
 from which equations we may deduce as above the equation 
 to the envelope, and the equations to its characteristic, and 
 edge of regression. 
 
 The tangent plane to the envelope at any point is also 
 the tangent plane to the generating cone, and therefore makes a 
 constant angle yy with the plane of wy; but sec y= V/l+p'+qs 
 therefore, since a=tan-y, the differential equation to the 
 surface is 
 
 p+g=a’; 
 the integral of which is of course represented by the system 
 
 of equations to the characteristic. 
 
 Hence, if S’ denote the area of any portion of the surface, 
 as! 
 dudy 
 
 that is, any portion of the surface bears a constant ratio 
 to its projection on the plane of wy. The characteristic in 
 this case is evidently a side of the generating cone, and 
 
 therefore perpendicular to the intersection of the tangent plane 
 with the plane of wy; therefore its two equations are 
 
 any +p +q= secry, or S = A, secy; 
 
 dy q 2. 2 2 
 da p’ Pq = 2s 
 
 hence the curve of greatest inclination of all surfaces generated 
 in this manner is a straight line, inclined at a constant angle 
 to the plane of vy. 
 
 9. The following are examples of envelopes when the 
 equation to the series of surfaces contains two parameters. 
 
 Ex. 1. To find the equation to the surface touched by 
 a series of planes so drawn, that the product of the per- 
 pendiculars let fall upon any one from two fixed points is 
 invariable. 
 
 Let the line joining the two points be the axis of «, 2a its 
 length, and the origin in its middle point; z=Ax+ By +e 
 
293 
 
 the equation to one of the planes; then the product of the 
 perpendiculars upon it will be found to be 
 ce — A’ a” : 
 eT ae b° suppose, or c’= 6°(1 + 4°+ B’) + A’a?. 
 
 Hence, differentiating the equation to the plane successively 
 with respect to A and B considering ¢ as a function of 
 A and B by virtue of the above equation, we have 
 
 A 
 i Ce) or ates GHA Se 
 Cc 
 
 B B : 
 OD alin 12 a or y = — b’—; and consequently x =— ; 
 r) Cc Cc 
 
 9 
 
 xv” . y i" 2 (P4+D)4+VP B+ ; 
 * a? +6 BB - Cc ey 
 
 the equation to a prolate spheroid, whose axis coincides with 
 
 the line joining the two given points. 
 
 Ex. 2. Let the equation to a surface be 
 
 ™m 
 
 v y” on 
 ay a Ys bn + 
 
 a” Pie 
 
 to find the equation to the surface which it touches in all the 
 positions it can assume, subject to the condition a”+ b"+c"=k" 
 a constant quantity. 
 
 Differentiating the equation to the surface successively with 
 respect to a and b, regarding c as a function of a and b by 
 virtue of the equation of condition, we have 
 
 a™ gm q"-1 a” gm a” 
 ake yas iC apemirian as Re er e e 
 y” 2m br-} y” g™ b” 
 m+1 m+l atts a 0, OF Fh Dn a 
 b c c Cor'C 
 Pare me Se Sa Ot ot et pee gen 
 ete Aa ae Toe TGR as xe Om A nt J ee 
 a b Cc c c AY»; k k 
 
294 
 
 mn mn mn 
 Z\ m+n Cc -nilarl U\ m+n a” y\ m+n b” 
 or y. = ie 5 similar af ie = ie 9 4 — ke 9 
 mn mn man mn 
 
 =, BM TB ymin 4 ymin — fe™+", the equation required. 
 
 If m=n=1, the equation becomes \/ 8 + / a + V/y = / kes 
 and agrees with Prob. 10, p. 238; ifm =n=2, itiszs+a@+y=hk, 
 which shews that if a series of surfaces of the second order 
 have the sum of the squares of their axes constant, they are 
 
 all touched by a plane. 
 
 Similarly, if the equation of condition be abe = k’*, it may 
 3 
 
 aye 
 be shewn that the equation to the envelope is vyx = (=) ; 
 
 and if m=1, we find wyz = Ss for the equation to the surface 
 
 touched by a series of planes which form with the co-ordinate 
 planes a pyramid of constant volume. ‘The surface repre- 
 sented by the equation wyx =a constant, has also the property 
 that the tangent plane at any point 2, y, x, forms with the 
 co-ordinate planes a pyramid of smaller volume than any 
 other plane drawn through the point of contact, the equation 
 to the tangent plane being 
 } , 
 moth Ld poe leg, 
 Ue Pm os 
 Kix. 3. To find the equation to the surface touched by a 
 series of planes which cut off from a given right cone, an 
 oblique cone, such that its volume is constant, or such that 
 the transverse axis of its elliptic base is constant. 
 
 If the volume of the oblique cone be constant, then by a 
 property of conic sections the transverse axis of the elliptic base 
 is also constant, suppose it’... = 2c, and let 6 = semi-vertical 
 angle of the cone; then it will be found that the equation to 
 the touched surface is z* = cot” (a + y’ +c’), the origin being 
 at the vertex ; which belongs to a hyperboloid of revolution 
 about the axis of the cone, and to which the surface of the 
 
295 
 
 cone is an asymptote, the 3 axes of the generating hyperbola 
 being c and ccot 6; also the point of contact is the center of 
 the elliptic section. 
 
 In like manner if a series of planes cut off from a parabo- 
 loid of revolution a segment of constant volume, the surface 
 touched by them is a similar and equal paraboloid about the 
 same axis, 
 
 Ex. 4. When from three straight lines meeting in a point 
 a plane cuts off three segments such that the product of two 
 of them divided by the third is constant, the plane constantly 
 touches a hyperbolic paraboloid. 
 
 Ex, 5. To find the equation to the surface which is 
 always touched by a plane whose equation is 
 me+ny+iz=v,,.,..(1), 
 
 m and 2 being independent parameters, and J and wv functions 
 of m and m determined by the equations 
 
 Differentiating (1) with respect to m and n, and taking 
 account of (2), we have 
 
 7) dv n dv 
 vex (-7 ree y+e(-7) ==. 
 
 But differentiating (3) with respect to m, and putting for 
 
 dv é 
 —— the value found above, we have, making 
 dm : 
 
 m n ia 
 eg bg toe A 
 (v° he a’)? (v" = b*)? (v" = c’)? > 
 
 1 1 & & 
 - Z of (e4- *) 3 
 al (hee a Co & ; 
 
 similarly, 2s) tae A ie = *) v; 
 
296 
 
 therefore, multiplying by m’ and n’, and adding, 
 
 m” n* P—1 ?—1 
 +5 = A (mr t+ ny + 
 Pe 0 
 
 o” —-q? a 
 
 4 ee z J y 
 a s—a- 4(F-0)o, apo 4(Z-0)e. 
 
 Therefore multiplying the squares of these three equations by 
 , m*, n”, and adding, 
 A= A (a +y' + 2° 4+v°—2v°)v* by (1) and (2), 
 
 or 1= A (7 — wv’) v*, making r? =a? + y' 4+ 2”. 
 
 pes lv Sm ay wt ea ra 
 Hence from (4) =—, VOCS 
 y— or gz—lu v~ zu — lv 
 
 and adding unity to each side of these equations we find 
 
 Yr —c Ps ce sv—lr 
 rP—v g—lv’ vw xv—lv?’ 
 aie aie C* % v (xv — Ir’) 
 therefore eliminating z — Jv, we have ———. = —____ ; 
 2 2 2 2 
 Yr —C r—v 
 . nilarl a? @ v (wv — mr’) bey v(yv —nr"’) 
 similiar => — = —_—___—_— ; 
 y 72 — QQ? yp? — a? y? — Bb? y2 — 
 
 therefore multiplying by z, x, y, respectively, and adding, we 
 have the required equation to the surface, 
 
 Cc gy? a” uv’ b? y” 
 
 v 
 
 2 2 
 + ——. + =" -,, = = ("0 - rv) = 0.~ 
 (ee | ga ar = 2 ae | ) 
 
 10. To find the singular points of the surface 
 
 a? x’ b? y Cc 3° 
 Pic? Cea Tae 
 or u=7"(a?a?+b°y’ +0") — a2a?(b’ +07) —b’y?(a? +0”) —c°s" (a? +b’) 40°’? =0, 
 
 a, b, and ¢ being in descending order of magnitude. 
 
297 
 
 We must have (Cor. Art. 108) ze = 0, ie = 0, = meri 
 or w {a2(r?— b’ - c?) + a? a’ + by’ + c?2°} = 0, 
 y {P(r — a? — oc) + a? a’ + Dy’ + c?s"t = 0, 
 sf{e(r-a@-B)+ aa +b y+ &x*t =0. 
 
 We can neither suppose the three variables to vanish together, 
 nor two of them to vanish together, for by neither of these 
 suppositions is the equation «=0 satisfied. Also if we 
 make w or x vanish separately, the values of the other co- 
 ordinates become Sat deh but if we make y = 0, we find 
 
 a? — b? Bae 
 vw=+c - ee Ea 
 - a’ — 
 
 which are real values and satisfy w=0; and at the point 
 corresponding to these values, we get 
 
 d? a—-B du 
 Bese Wits GG RAY Sima saountrse 200) 
 a’ sate? BP—c du 
 ——— | C. SSS SS 
 d 3° a’—c? dady ’ 
 au a? + ¢ = a eae EE ea 
 SI hens ah ie te VG . = Ponca)» S 
 
 consequently the equation to the tangent cone at the singular 
 point, taking that point as origin, becomes 
 
 2 
 uv? ae, a a +c os 
 
 Boe ace 7 "PP * /(@- b?)(b?-c) ac 
 
 Also for the distance of the singular point from the center 
 we have, substituting its co-ordinates in (1), 
 
 2-h R-@& 
 
 + : 
 Poee gt FS 
 
 =0; which gives 7? = b’, 
 
 If fig. 60 represent the intersecting ellipse and circle that 
 form the principal section of the surface in the plane of az, 
 then the intersections of those curves, one of which is R, are 
 
298 
 
 the singular points, and the two sheets of the surface meet 
 in them. 
 
 11. To shew that tangent planes can be found which shall 
 touch the surface in the preceding Problem in circles whose 
 
 GEA) 
 
 Ae 
 radll ob 
 
 The section by «z is an ellipse and a circle intersecting one 
 another represented by the equations 
 
 (a +2 — b’) (a?x 4 Ce? ac?) = 0 (fig. 60). 
 
 Let PQT be a common tangent to this ellipse and circle, 
 then CQ the perpendicular upon it = 6, let zQ7'C = 0, then 
 
 2 
 CT =bcosec@, T7'Q= bcotd, TP = — cot 6, and cot@ 
 
 6 oe Cc e ee 
 = yf ——,,- We shall begin by determining the intersection 
 a-— 
 
 with the surface of a plane perpendicular to s@ passing 
 through the common tangent PQT'; for which purpose we 
 must, in the equation to the surface, change 2 into — x cos 9 
 + bcosec @, and x into wsin @ (Art. 101) which gives, dividing 
 by 6° and observing that a” cos’ 0 + c’ sin’ 0 = BD’, 
 
 2a? 
 (a + y? + cosec’?@ — 2bacot) (a +4" +a’ cosec’™ — yp weot 0) 
 
 2 2 
 a’c 
 —2° (2 + 
 
 ) yeye 
 
 2a” 
 “t oe ( +c’) a cot 8 — ab’ cosec® 9 — a*c’cot?@ = 0. 
 
 From this we get by successive reductions, taking notice of the 
 value of cot 0, 
 
 g 
 (a +)? + (a +’) {G + b°) cosec” 9 — 2x cot 0 (2 oe =) 
 
 + a°b’ cosec’O — 4.2.a7b cot @cosec?@ + 4a°x* cot’ @ — (x* + y*)(a" +c’) 
 2 
 
 + x°(a°—b") (1 ae, 
 
 207 2 . 
 7a) one (U' +c") cotO—a’b’cosec®8 ~a*c’cot*O=0, 
 
299 
 or (wv +9’)? + 2(a° + y’)acot 6 {0 cot 0 — & (< ee -)| 
 a 
 ek egy. ey ay: 
 + (a cot 0)* — 2(acot 6)° (5 ae -) v® + aa’ cot’@ ¢ ae = 
 a a 
 
 or | a’+y? , a AL 
 v+y +tacot@iacotd—wx ae = 0, 
 a 
 
 which shews that for the assumed position of the cutting plane, 
 the two curves of intersection become united in a circle (and 
 consequently the plane touches the surface at every point in 
 the circle) whose equation is 
 
 a Ob 
 y? = acot@ (; + -) 2 — a’ cot?@ — 2” 
 
 2 
 
 “= ( cot 8 — v) (@ — bcot 0) =(PT —- x)(« - QT), 
 
 so that PQ is the diameter of the plane circle of contact; also 
 
 22 
 by the ellipse CP’ = a’ +c? — — 
 2A 
 1 
 . P@ =a - B+ 8 - ae =a (a - BY - e). 
 
 ProspLeMs on: Section IX. 
 
 1. Asa first example of this theory, take the curve of 
 double curvature, resulting from the intersection of a sphere 
 and cylinder, considered in Prob. 1, p. 253. 
 
 Its equations are y°=2av—a"*, x°=4a°—2aa, and the 
 equation to the normal plane, as we have seen, is 
 
 To obtain the equations to the line in which this is inter- 
 
300 
 
 sected by the consecutive normal plane, we must join to it 
 the derived equation with respect to #, namely 
 
 1 a-«a« dy a dz 
 u(-f- GB) EE 
 
 or, by substitution, 2 +4 — =0,......(2)5 
 fois 
 and to find the equation to the surface generated by the per- 
 petual intersection of the normal planes, we must eliminate 
 v, y, and x, between these two equations, and the equations 
 to the curve. 
 Yuen Pee (ve 
 Now from (2) (-%) == or — =2(%) : 
 x x 
 
 3 
 , 2a a 
 
 . v a—- @ 
 and (1) may be written — ey hiss eee 
 x %, a x 
 
 é 
 nq Nal uagip eae a 
 & av Xv Xv 3 
 or — =(2-2) +%(1-=) + (%) = 0. 
 x a a % a} x 
 é 4 4 
 e e v e e 
 Hence substituting for — its value, and reducing, we find for 
 a 
 
 the surface generated by the intersection of the normal planes 
 
 2 eT ET Te, 2 ee 
 2(e, - yp/ 23 -y)= 8/85 — y 3. 
 
 a 
 Al 1 - 2 2g 
 sO +y° +a ay | a +2) 
 Ad dA ") eel A ih de a a 
 y+ @')+ y's ~2"y)'= (F5) (loa +52) 5 
 (2a 4+ 2x)? 
 
 .. radius of curvature = —7—="—.. 
 \/ 100+ 3H 
 
 2. On a given surface, to trace a curve of such a nature, 
 that its involute shall be a plane curve. 
 
 The curve of constant inclination, that is, one whose tan- 
 gent line is inclined at a constant angle to the plane of ay, will 
 have this property. For if s denote the length of the are, inter- 
 
301 
 
 cepted between the plane of wy and a point whose ordinate 
 is x, and + = the constant angle which the tangent line makes 
 
 f ; dz 
 with the axis of x, then rF = COS y ; 
 Ss 
 
 *. S=scosy, ands =~2 secy 
 
 = the length of the tangent line between the point of con- 
 tact and plane of wy; hence if a thread be applied to the 
 curve of equal inclination, and then be unwound from it 
 beginning from the plane of wy and be kept stretched in 
 the direction of a tangent, the extremity of the thread will be 
 always in the plane of wy, and will describe a curve which is 
 manifestly the involute of the projection of the curve of equal 
 inclination on the same plane; for the projection of the tangent 
 line will touch the projection of the curve, and will be of the 
 same length. The equations to the curve of equal inclination 
 were found at page 257. 
 
 Ex. Let the surface be a right cone and the origin in~ 
 its vertex, so that p = x tana, 
 
 2 Fe eA oes Ue, 
 cotta = cot! y 5» or Pp =pV/1- (tana cot y)’, 
 
 the equation to an equiangular spiral. Hence it follows that 
 this curve cuts the generating line of the cone under a con- 
 stant angle, that is, it is the conical helix. Hence if a thread 
 be applied to the surface of a cone, according to a helix traced 
 on it, and then be unwound beginning from the vertex, its 
 extremity will always be found in a plane perpendicular to the 
 axis through the vertex, and will trace out an equiangular 
 spiral on that plane, which is also the involute of the pro- 
 jection of the conical helix on the same plane, and therefore 
 similar to it. 
 
 3. If a uniform and flexible string be suspended from 
 two points in the surface of a vertical cylinder whose base is a 
 circle, it will form itself into a curve of double curvature, such 
 that the involute lies on the surface of a sphere. 
 
302 
 
 Let A (fig. 61) be the lowest point of the catenary, 4B 
 the radius of curvature at that point; take the horizontal plane 
 drawn through B for the plane of wy, and the axis of the 
 cylinder for that of z, and let CM, MN, NP be the co- 
 ordinates of any point P in the curve. Draw NT" per- 
 pendicular to the tangent PT’ and join C7’; then PT is 
 perpendicular to the plane C7'N, for a line drawn through NV 
 parallel to PT’, and therefore coinciding with the tangent 
 plane of the cylinder would be at right angles both to 7'N, 
 CN, and therefore to the plane passing through them; and 
 CNT is a right angle, therefore 
 
 CT? = CN’ + NT® = CB’ + BA’ = CA’, 
 
 because the perpendicular upon the tangent from the foot of 
 the ordinate, is constant by the nature of the catenary. Hence 
 the locus of 7’ is on a sphere radius CA, and since CT'P is a 
 right angle, therefore PT' touches the sphere. But PT'=arc AP; 
 if therefore a thread be unwound from AP beginning from A, 
 and be always kept stretched in the direction of a tangent, its 
 extremity will trace out a curve situated on a sphere and the 
 thread itself will be a tangent to the sphere. 
 
 4. In a curve of double curvature, if S'Y be a perpen- 
 dicular on the tangent at any point P from a fixed point S, 
 and SY" be a perpendicular on YY’ the tangent to the locus 
 of Y, then, 2 being the perpendicular from S$ on the plane 
 through YY’ and the middle point of SP, 
 
 SY! = SP’ (SY? -— h’). 
 
 In fig. 55 bis take Q, R, points of the curve on opposite 
 sides of P and very near to it, and let qg, 7, be the points of 
 intersection of the perpendiculars from §' on the chords PQ, 
 PR produced; these points lie on a sphere diameter SP. 
 Through O the middle point of SP, and through the line qr 
 draw a plane cutting this sphere in the circle gYr. Now let 
 Q; R, move up to and coincide with P; then Pq, Pr both 
 coincide with PY the tangent to the curve at P, and qr 
 coincides with YY’ the tangent to the locus of Y; and SY’ 
 
303 
 
 being the perpendicular from the vertex of a cone upon a tan- 
 gent to the circular base, it is easily shewn that 
 
 SY‘ = SP*. (SY? —i’). 
 
 Hence the normal plane to the locus of the foot of the per- 
 pendicular on the tangent to any curve from the pole, always 
 bisects the corresponding radius vector. 
 
 ProBLEMS ON SECTION X. 
 
 WE shall next apply the results obtained in Section X. to 
 several particular cases. 
 
 1. To find the equation to a plane which shall have a 
 contact of the first order with a surface at a proposed point. 
 
 Since the equation to a plane (which here takes the place 
 of V=0, Art. 212) contains only three arbitrary constants, 
 ‘a plane can in general have only a contact of the first order 
 with a surface completely given. We have therefore 
 
 s=Av+By+c, P=A, Q=B, 
 
 from which there results between the constants A, B, c, and 
 the co-ordinates x, y, s of the proposed point in the surface, 
 the relations s =Axv+ By+c, A=p, B=q; subtracting 
 therefore the value of x from that of x’, and putting for A 
 and B their values, the equation to the plane having a contact 
 of the first order with a given surface at a point vyz, the same 
 as that to the tangent plane, is 
 
 s—#= p(w —ax)+q(y'—y). 
 2. The equation to a sphere is 
 (2 - a)’ + (y' - BY’ + (@ - 7) =, 
 and since it contains only four arbitrary constants, a sphere 
 cannot generally have a complete contact of the second order 
 
 with a surface. Suppose it required only to have a contact 
 of the first order ; 
 
304 
 
 then ia SE, inne? 
 dx s—y dy em—+ 
 
 Hence there results between the constants a, 3, y, 6, and the 
 co-ordinates wv, y, %, of the proposed point of the surface the 
 following relations, 
 
 (7 —a)*+(y — B)?+ (2 -y)P= oe, — 
 
 G—-a y-p 
 — P> —_— —___ = 
 e-¥ s—¥ 
 or w-a+tp(e-—y)=0, y-B+q(e- +) =9, 
 which equations shew that the centers of all the spheres lie in 
 the normal to the point of contact. 
 
 q> 
 
 r) 
 Also (z- iF l+p+q = 0, ors— ee 
 mt z Vit pre 
 po qo 
 and az=@&#-4+ SS 9 Bp=y+— ———.. 
 ASE rig \/ Le rae 
 
 Hence the number of spheres, which may have a simple 
 contact with a surface at the same point, is unlimited; the 
 radius of the sphere being assigned, the co-ordinates of its 
 center are given by the above equations. 
 
 3. To find the radius of curvature of any normal 
 section, at a given point of an oblate spheroid. (Art. 229.) 
 
 Let /= 2 PGA, PG being a normal (fig. 67); and let the 
 
 section PQ make an angle a with the meridian, and its radius 
 of curvature = R; 
 ; a a(l-eé 
 ‘+, radius of curvature of meridian PO = p = eed : 
 (1 — e’ sin?/)3 
 
 c a 
 radius of curv. perp. to meridian PO’ = p’ LIP 
 1 — e’ sin® 
 
 pp. a 1-é 
 iz p sin’a + p cos’a oY Ay esin?] 1 ~ e? + e’cos’acos*] 
 4. To find the umbilici of an ellipsoid. 
 These are points (Art. 226) at which, if the ellipsoid were 
 
305 a 
 
 generated by a circle of variable radius moving parallel to 
 itself, the plane of the generating circle would become the 
 tangent plane. Let M (fig. 67) be one of these points which 
 is necessarily in the principal section containing the greatest 
 and least axes, and let AZ7' be the tangent at MW; then 
 
 c a—b ec av 
 tanta / Oo 
 F oe GPa = (Art. 104) 
 
 of 7 = A c 9 6 
 if CN =a, since the equation to BPM is x = —\/a?— a’; 
 a 
 
 hence the co-ordinates of M, and of the three other umbilici 
 similarly situated in the other quadrants of the principal 
 section, are 
 
 a — b amy 
 e=+ta ‘ 2 0° z<=tec 
 — C™ 
 
 a a” — 
 
 Also the radius of curvature at an umbilicus 
 
 since it is the radius of curvature of the principal section cor- 
 responding to these co-ordinates. 
 
 5. To determine the lines of curvature of an ellipsoid. 
 
 9 
 ~~ 
 
 A Wik 
 Let the equation be —+5 4 —-—=1; 
 +3 
 
 dx y v ne yy 
 
 poly dp az OY chi PB 'sF 
 
 dp ne 2 dp a 
 
 ate Ora y)3"8 dy ape 
 
 OG, sep ty MY 2» 
 ERIE Na ee 
 
 therefore, by substitution in the equation of Art. 231, 
 
 a(B-y) avy eS) 
 20 
 
306 
 
 + {B(a-~7)w*—a(B-y)¥"-aB(a-B)} —" - Bla-y) ay =o, 
 
 dy\* dy 
 A sonrs 2 — Ay*® — B) — -: =O: 
 eae ey) gre : De + ? 
 tate) Sig Gk ea 
 B(a-) Cary 
 
 which will be both positive if we suppose the plane of pro- 
 jection, that is, the plane of wy to contain the greatest and 
 the mean axis, so that a, 8, y are in descending order of 
 magnitude. 
 
 calling > 
 
 ; ; { d LU > 
 To integrate this equation, let d =» UE being a 
 v Yy 
 
 function of v; 
 
 x? 2 
 
 * vy (4 3 - 1) +5" (@*- dy'- B) =o, 
 
 or Aa’ue—y’?+u(a’? —-Ay’® — B)=0, 
 
 Bu 
 o P(l = au (1 _ 1 OF i= oe 
 y(1+ Au) =a?u(1+ dAu)— Bu, or y=2°u my 
 du B du 
 
 : tent d 
 Hence differentiating 2y = 2UUu + x ain Guna 7? 
 r ek Fe i 0, since 2 dy g 
 OC e — = ; 
 (1+ Auyf ’ Scar oa 
 
 . oa : : of 2° 
 . u =C, which gives y c (a a) 
 
 the complete integral, C being the arbitrary constant. 
 
 Also taking the other factor # = j , we have 
 
 + Au 
 y? =u (0° - #\/B) = 7 (@- VB) (VB -2), 
 or (yA) + (@- VB) =0 
 
307 
 
 the singular solution, which resolves itself into the two 
 
 mir ee BP 
 y =0, Pa aly Sora asn ale 
 a — Cc 
 and indicates the umbilici of the surface (Prob. IV.), the normals 
 at which, as we know, are intersected by the normals drawn 
 from consecutive points in every direction on the surface. 
 
 6. To construct the projections of the lines of curvature 
 
 of an ellipsoid upon the plane containing the greatest and the 
 mean axis. 
 
 First, to determine the constant in the complete integral, 
 let a, y’ be the given co-ordinates of the point through which 
 the lines of curvature pass, 
 
 A ey B Ayiiti Bi whol gh 
 te Meaeee 10) oe Cais tases Sa 
 Ay?+B— ax 1 Aa LI ee Li a 
 oe a pag Ave + BE ah aday 
 Ay” + B+x? ; 
 and Lig hea sre 3G? (Ay? +B +a) — 4Ba”, 
 
 Hence, we see that C has two real values one +, and the 
 other —, and that 1 + AC is always +; hence the complete 
 integral represents an ellipse when C is —, and a hyperbola 
 when C is +; and therefore the projections of the lines of 
 curvature passing through any point on the surface of an 
 ellipsoid upon the principal plane containing the greatest 
 and mean axes, are an ellipse and hyperbola. 
 
 In the first case, if m and m be the semi-axes of the ellipse, 
 
 is the equation by which m and m are connected; in the 
 second case, if m and m be the semi-axes of the hyperbola, 
 n* : B Win Ae 
 
 Wee AC? Bt BS 
 20—2 
 
308 
 
 Hence the semi-axes for each ellipse in the projection are 
 the co-ordinates of a point in the same given hyperbola, and 
 for each hyperbola the semi-axes are the co-ordinates of a 
 point in the same given ellipse; these are called the auxiliary 
 ellipse and hyperbola; they are concentric with the ellipsoid, 
 and their semi-axes major and minor (which are the same for 
 both curves, and coincide in direction with those of the 
 ellipsoid) have the following values 
 
 SJB =o oa ana M/E nent 
 
 a Cc” a 
 
 Hence we have the following construction for the pro- 
 jection, upon the plane containing the greatest and mean 
 axes, of the lines of curvature of an ellipsoid. Let ACB 
 (fig. 68) be this plane (that of wy), and with the axes 
 found above construct the auxiliary ellipse and hyperbola 
 GO, HO. From any point J in the hyperbola, and 7 in 
 the ellipse, let fall perpendiculars on the axes CA, CB; and 
 with semi-axes CN, CM construct the ellipse JZN, this 
 is the projection of a line of the first curvature; and in the 
 same manner the projections of any number of lines of the 
 first curvature may be constructed. Again with semi-axes 
 mi, ni construct the hyperbola »X, this is the projection 
 of a line of the second curvature; and similarly any number 
 of projections of lines of the second curvature may be con- 
 structed. 
 
 As the points J and i approach O and G, the ellipse and 
 hyperbola continually approach the lines CO, CB, and are at 
 last confounded with them; therefore CO, CB are the pro- 
 jections of lines of curvature, and consequently the curves in 
 which the surface is cut by the principal planes of wz, yx are 
 lines of curvature; also if in the equation to the auxiliary 
 hyperbola we make m = a, we find n = b, therefore the ellipse 
 AB is included in the construction, and is itself a line of 
 curvature. Hence the intersections of the surface with its 
 three principal planes are all lines of curvature. Since CO 
 
 ~ Bb 
 a= wih 72 , and is therefore <a, the point O falls within 
 
 a 
 
309 
 
 the ellipse 4B; it is the projection of an umbilicus of the 
 surface (Art. 235) round which the lines of both curvatures are 
 symmetrically disposed, and towards which they all turn their 
 concavities. 
 
 If the plane of projection contain the mean and least axes, 
 i.e. if -y, B, a be in descending order of magnitude, then 
 
 APRA EC. of Ne y-B 
 
 are still both positive, so that this 
 
 case is resolved into the preceding. 
 
 7. To construct the projections of the lines of curva- 
 ture of an ellipsoid upon the plane containing the greatest 
 and least axes. 
 
 If we suppose the plane of projection, that is, the plane 
 of xy to contain the greatest and least axes of the ellipsoid, 
 we shall find the construction easier, In that case a, y, 3 
 are in descending order of magnitude, and 
 
 suppose; B still remains positive. Hence 
 
 B 
 Wiens (Jab pawensit os fee k : 
 y (« Fa) 
 
 {Ayy? - Bra? &/ (Ay?- B+ay—4Ayay?}. 
 
 Hence both values of C have the same sign, and are 
 both negative 
 Ayy” a’? 
 
 B Lent 4. e. if 
 
 re 
 
 — 2 Aw? 
 
 {Z 12 Ae 2 
 fies 219 ait Cen ene 
 a- 23 a-Ba 
 
 which must always be the case, because the given point being 
 (2 tela 
 sah Hae 1. 
 
 eee a 
 
 if 
 
 in the surface of the ellipsoid, we must have 
 
310 
 
 Hence both projections are ellipses; and if m and n be 
 the semi-axes of any ellipse, we have 
 
 n? C. om? B m An? 
 a. es n= —_ Aga 
 a Peri er Rawr 
 which is the equation to an ellipse whose semi-axes are 
 
 ome a? — Bb 7 fone 
 VAPELVA ond V2 at olor 
 
 a—-¢ 
 
 In this case, therefore, the semi-axes of each ellipse of 
 the projection are the co-ordinates of a point taken on another 
 given ellipse, which is the same for all. 
 
 Hence let BA (fig. 69) be the principal section of the 
 ellipsoid containing the greatest and least axes; and concen- 
 tric with it describe the auxiliary ellipse Y.Y with the semi- 
 axes just found, and which we observe are greater than the 
 corresponding semi-axes of the ellipsoid @ and 6. From any 
 points J, i let fall perpendiculars on OX, OY and with these 
 as semi-axes construct the ellipses MN, mn; then MN, mn 
 are the projections of lines of curvature. If in the equation 
 to the auxiliary ellipse we make m =a we find n = J, therefore 
 the principal section is itself included in the construction; and 
 if from the points 4 and B we draw tangents meeting the 
 auxiliary ellipse in D, the perpendiculars dropped from every 
 point in DY will serve to construct the projections of lines of 
 the first curvature, and perpendiculars dropped from the points 
 in DX will serve for the construction of the projections of 
 lines of the second curvature. 
 
 It is not difficult to shew that if AY be joined, all the 
 ellipses in the projection are touched by it, and consequently 
 they are all inscribed in the same equilateral parallelogram. 
 This however must appear from the singular solution of the 
 differential equation to the lines of curvature, for that re- 
 presents a curve which always touches and envelopes the curves 
 
 rized in the complete int 1. Now the factor v?— ———_— 
 comp omplete integra Oo (14 dat 
 
311 
 
 B 
 (p. 306) which in this case becomes a — G-Auye? put equal 
 
 to zero, gives a value of w which substituted in the equation 
 
 u 1 a 
 iv at eee (ee B); 
 gives ¥ r, 6 VB) 
 
 2 = wu — ——_— 
 y 1 — A, 
 
 and this equation represents the four chords which pass 
 through the vertices of the auxiliary ellipse, and of which YX 
 is one touching AB at an umbilicus of the surface. 
 
 The conclusion from the whole investigation is, that if the 
 two lines of curvature which pass through any point on the 
 surface of an ellipsoid be projected upon the principal planes, 
 the projections will be an ellipse and hyperbola, or two ellipses, 
 according as the plane of projection does or does not contain 
 the mean axis of the ellipsoid. 
 
 Cor. 1. Since the equations of condition which charac- 
 terize umbilici always make the differential equation of the 
 lines of curvature identical, and since that equation in the 
 present case is 
 
 dy\? dy 
 A om *_ Ay’ — B)— -—wy=0......(1), 
 vy (52) + (@ ig Bisa ey (1) 
 we must have wy=0, a? — Ay? — B=0; but we cannot sup- 
 pose w=0, for that gives y® equal to a negative quantity ; 
 therefore we must have 
 2 BP 
 y=0, «= 2B, or «= aL! 
 
 6 » 9 
 a? — Cc’ 
 
 which equations indicate four points placed symmetrically in 
 the four angles of the principal section containing the greatest 
 and least axes, and of two of which O (fig. 68) is the pro- 
 jection. ‘These expressions for the co-ordinates of the um- 
 bilici agree with those previously found by considering that 
 they are situated at the extremities of the diameters which 
 pass through the centers of all the circular sections of an 
 ellipsoid. 
 
312 
 
 Cor. 2. In constructing the projections of the lines of 
 curvature of an ellipsoid upon the plane containing the 
 greatest and the mean axis, we saw that for each of the 
 umbilici the lines of curvature passing through it were re- 
 duced to a single one, viz. the vertical ellipse projected 
 into COA. ‘This also appears from the differential equation 
 
 cs e d e 
 to the lines of curvature; for since oe for the particular 
 wv 
 values of w and y belonging to an umbilicus, assumes the 
 O ae? : 
 form Bere must, to obtain its true value in that case, recur 
 to the next derived equation according to the principles laid 
 
 down in Art. 235; this gives, not writing down the terms 
 2 
 
 which involve —~ 
 
 dx?’ 
 dy\* dy\* dy 
 Aw - _— p>“ -—y=0; 
 = (=!) Ay (=) pee dx J 
 
 but at an umbilicus when y= 0, a= £1/B this equation is 
 
 decomposed into 
 if wv dx 
 
 is the only admissible value, which when integrated and cor- 
 rected so as to pass through an umbilicus, gives y=0, and 
 indicates the vertical ellipse which is projected into COA. 
 
 d dy\? d 
 an, 4 (=) +1=0; therefore = 0 
 
 8. The lines of curvature of an ellipsoid are the curves 
 in which it is intersected by homofocal surfaces of the second 
 order. 
 
 Let the equations to a given ellipsoid and to another 
 homofocal surface be 
 
 2 2 2 z 2 
 
 eh ap tis PMA Dee 
 mg Ta ta Oe Gaolegad pam de) 
 
 then the plane of wy being supposed to contain the greatest 
 and least axes of the ellipsoid, the constants a”, b%, c”® are 
 subject to the conditions 
 
 2 2yeee 2 Ae 2 PE Pak. ae er 2 Qian 12 a 
 a baa." — b*, a —~c= a4? — 07, 1c bv c* —bee) 
 
313 
 
 which amount only to two distinct equations, so that one of 
 those constants is arbitrary. ‘The co-ordinates of all the points 
 of intersection of (1) and (2) by Prob. VI., p. 237, satisfy the 
 equation 
 
 ae y” 3? 
 
 eat ppt poe? 
 which shews that some of the constants a”, 6”, c’® are negative; 
 and from equations (3) it must be either 6” alone, or b” and c¢” 
 together that are negative. Hence the curve of intersection 
 of (1) and (4) which is the same thing as that of (1) and (2) 
 
 has for its projection the equation 
 
 @ 
 ; art peaiom 2 e-—-B)=1, (5 
 on J (8-8) =1, ©) 
 which represents an ellipse since 6” is always negative. Let 
 X and Y denote the semi-axes of this ellipse, then 
 
 2 2 
 ; Bae eck =a? — b, 
 a” b? 
 
 so that the semi-axes of (5) are subject to the same condition as 
 are the semi-axes of the projections of the lines of curvature of 
 
 the ellipsoid by Prob. VII.; consequently the curves in which 
 
 2 2 2 
 
 o ~ ~ 
 
 e ° e e v e ° 6 . 
 the ellipsoid is intersected by — — 7 + —; = 1, in which a” is 
 oe Yas 
 
 arbitrary but 6” and c” subject to the conditions a” + 6” = a? — b’, 
 
 a® —¢c® =a’ —c’, are its first lines of curvature. Similarly the 
 2 2 2 
 . ° . dh J] S$ 
 
 second lines of curvature lie in the surface — — co oe 1; 
 
 where a” is arbitrary, but 6” and c” subject to the conditions 
 a’ 4b? =a?—0, a? +c* =a’ —c’. And equation (4) shews 
 that those lines of curvature are likewise situated on a conical 
 surface of the second order concentric with the ellipsoid. 
 
314 
 
 9. To determine the radii of curvature of an ellipsoid. 
 
 . of ies : 
 Let the equation be — + B 4+ —=1,then taking the values of 
 a w 
 
 the differential coefficients from Ex. 5, we have 
 ’ 2 2 2 
 ee) t= — x (~) (=) _# 
 G4 pyt=- laa (2) (E) @- a}, 
 ry) (x)' aaa 
 oe appa acd A 2? a BY lc 9 
 ae ra «)° apt 
 
 spate #+(2) (f) @- 99}, 
 
 But in taking the sum of these three equations, the co- 
 
 2 
 efficient of (~) within the brackets is 
 g 
 
 a are 
 
 S+5- (Sef) (C48) (.-£-8)-o off) 
 (+ pi 2pqs+ + eee = 
 
 Pa fetBry- Way} =o 
 
 also rt — s? = (45) {(B-y)(a- 4) - wy} = : oe 
 
 a“ y 1 
 md taped S (54 B +3) yee 
 D being the distance of wy from the center, P the perpen- 
 dicular from the center on the tangent plane, and 4?=a+6+ y3 
 
 therefore by substitution in the equation of Art. 233 and re- 
 duction we have 
 
 (4 @- x)? + tf - D*) (z’ — #) + + Be = 0. 
 
 , 
 But Rea~(/-) /ispaeg = -2=9%, 
 z 
 
315 
 
 A’ — D’ ary 
 ag 3) ageieaie 2 = 
 kt P R+ pe = 0 (1) 
 Let R’, R” denote the values of # in this equation, then 
 A2 — D? 
 P ? 
 
 abe\? 
 - tore 2 , Reg 
 R.R (=) h+ 
 
 which is the simplest form to which the result can be reduced ; 
 the first equation shews that at points where the tangent planes 
 are equidistant from the center, the product of the radii is 
 constant; and the second that at points which are themselves 
 equidistant from the center, the sum of the radii varies inversely 
 as the perpendicular on the tangent plane. Hence the volume 
 of the ellipsoid may be expressed by the two radii of curvature 
 at any point, and the perpendicular on the tangent plane, for it 
 4cabe 
 
 _4 7pm ./PR 
 SSS =— PS RR’. 
 
 Also the radius of curvature at an umbilicus (since the two 
 radii become equal to one another) 
 
 b 2 2 Gb? 
 pein (= +5) = —, as before. 
 Pew iy ac 
 
 The equation (1) for determining the two values of R may 
 be presented under a different shape, leading to several im- 
 portant results; and may be investigated by the following 
 direct method that is equally applicable to all surfaces whose 
 equations are of a particular form. 
 
 10. To find the two radii of curvature at any point of 
 a surface whose equation is of the form 
 
 f (#7) +) +x) = 0. 
 
 In this case the quantities w’, v’, w’ vanish in the formula 
 of Art. 217, and we have 
 
 ‘ 
 
 TA APE Ss 1 oie 
 Pu+tmv +n'w = ous + V* + W? = P suppose, 
 
316 
 
 which must be a maximum or minimum by the variations of 
 l, m, m under the conditions (Art. 24) 
 
 Pamin?=1, 1U+mV+nWe=0, (2) 
 
 by virtue of which m and m may be considered as functions 
 of 7; hence differentiating the above three equations with 
 respect to 7, multiplying the two latter by indeterminate 
 coefficients } and «, adding them to the former and then 
 
 d dn 
 making the coefficients of a and Pr vanish, we get for 
 
 finding J, m, m the equations 
 lu+«U +aAl =0 
 mv +«V+Am=0> which give P+rA=0; 
 nmw+kW+rAn =0 
 
 =P o— PP 0 — P 
 ‘ob pa, es 
 
 U Ve W 
 
 consequently from (2) we get the equation furnishing p, p’ the 
 maximum and minimum value of & 
 
 : 1 Seo 
 (since P = aU + V* + W*), 
 
 U° V? Ww? i 
 ht gaa Oar eres a Te 
 x’ y° 3° 
 In the case of the ellipsoid — + = + —=1, this becomes 
 ham ao eey 
 x y° Prd 
 
 BGeiRp) a PCB) ay CReaR gy a 
 
 p being the perpendicular on the tangent plane at wyz. 
 
 We shall next deduce some results both with regard to 
 the radii of curvature and the lines of curvature of an ellipsoid, 
 from applying the general equation to lines of curvature given 
 in Cor. Art. 231, to that surface. 
 
. @& 
 
 317 
 
 11. To find the equations to the straight lines which 
 touch the lines of curvature at any point of an ellipsoid. 
 
 The equation of Cor. Art. 231 when applied to the ellipsoid 
 
 2 2 2 
 
 —+>+—=1 gives (considering each co-ordinate as a func- 
 ¥4 
 
 a 
 
 ‘ fer Gee 
 tion of a new variable ¢ and making sates: &c.) 
 
 y¥x(y—- Beta (a-y)y+ay(B-a)x=0, 
 
 yy ee’ 
 
 casera 
 
 Now the equations to the tangent line at vy are (Art. 124) 
 OY — ye Lee 
 
 ap eee Se pas 
 a’ y’ 2! 
 
 , 
 to which must be joined SETS as 
 a 
 
 therefore eliminating w’, y’, ’, the tangent lines to the lines 
 of curvature at a point wy are determined by the intersections 
 of the cone and plane 
 
 (Y-y)(Z-32)(y—- B)vt+(X-2)(Z-2z)(a-y)y 
 +(X-«)(Y-y) (6 -a)%=0, 
 
 (¥-«)"4(¥-y) 54 (Z-s)==0, 
 a B ye 
 
 the cone having its vertex in the proposed point, and the plane 
 being the tangent plane to the surface at that point. 
 
 12. The tangents to the lines of curvature at any point 
 of an ellipsoid are parallel to the axes of the diametral section 
 made by a plane parallel to the tangent plane at that point. 
 
 OR diy? > 3? 
 Detect B + — =1 be the equation to the surface, ryz 
 
 a 5. 
 tide sie. oC ele ZZ j 
 the point in it, and —- +— ee the equation to the 
 a 
 
 cutting plane; then r= X* + Y?+Z? is to be maximum, 
 
318 
 
 X, Y, Z being taken so as to satisfy each of these equations ; 
 therefore proceeding as in Ex. 10, 
 
 (a+A)X+Kv=0, (B+A)V+ny=0, (y+A)Z+nx=0, (1) 
 and eliminating \ and « we get 
 YZ(y—-P)v7+XZ(a-y)y+ XY (B-a)x=0; 
 
 ; woaekw | oY: ZZ t ; : 
 which with — + nes +—  =0 determine two straight lines 
 
 a 
 
 6 
 which are the axes of the diametral section; and these lines 
 are evidently parallel to the tangents to the lines of curvature 
 determined in the preceding Problem. Also for the magnitude 
 of the axes we get from equations (1) 
 
 7+rA=0, < (a=) = B=) =ly-0)3 
 
 ERE coe it, Uolss y y a ee 
 “a(a-r) B(B-1r) yy-n ” 
 
 which determines ”. 
 
 13. The radius of curvature of any normal section of an 
 ellipsoid = D’ +p, D being the semidiameter of the surface 
 which is parallel to the tangent line of the normal section, 
 and p the perpendicular from the center on the tangent plane. 
 
 For comparing the above equation for finding the axes of the 
 diametral section, with that for finding the radii of curvature 
 
 in Problem X. we have #p =r, which shews that the radii of 
 2 2 
 curvature at any point of an ellipsoid are equal to aa and —, 
 
 p 
 p being the perpendicular on the tangent plane at that point, 
 
 and 4A and B the semiaxes of the diametral section parallel to 
 that plane. Also for the radius of curvature R of any normal 
 section at the same point inclined at an angle @ to a principal 
 section, we shall have (Art. 222) 
 
 1 (cos’@ sin’ @ p 
 rae e* =) a 
 
319 
 
 14. For different points taken along the same line of 
 curvature of an ellipsoid, the product of the diameter of the 
 surface that is parallel to the direction of the line of curvature 
 at any point, and the perpendicular from the center on the 
 tangent plane at the same point, is invariable. 
 
 As in Problem XI., the equations to the lines of curva- 
 ture are 
 
 a (sy'— ys’) a + B (wa! — xa’) y+ y (ya’— wy’) x'=0, (1) 
 ae yi ee 0. 
 Ce Tas ame cy 
 Hence eliminating by cross multiplication, and taking » to 
 denote an arbitrary multiplier, we get 
 
 a’ = B (#2 - Bees ie y (yx - vy) 5 
 
 = Boye (“Ss het =)- Bryw (= +e +), 
 
 2 2 2 
 or putting = + ie a = =u, and \=pn.aBy we get for a’, 
 o res 
 and similarly for y’ and 2’, 
 x a old Baye dhs ore/’ par x 
 
 v=—uU——U y¥=—-t -—-—4& s=—uw—-—UuU, 2 
 lu Pte a ad a eg aca (2) 
 Hence multiplying these equations respectively by wv”, y”, x 
 and adding, and supposing the arc to be independent variable 
 
 EST 
 
 so that a’ a” + y’y” + ’3" = 0, and observing that 
 
 “v 
 
320 
 
 or PD = C as in Prob. VII. p. 289. Hence all properties de- 
 duced from this equation for the geodesic line of the ellipsoid, 
 will hold also for the lines of curvature. 
 
 Also this process leads to the integral of (1) and conse- 
 quently to the determination of the lines of curvature. For 
 equations (2) give 
 
 , , : , 
 VU ; u SU 
 a 9 2Y mB en! 3 Of a 
 
 uU+ ma uw+uld Ut ey 
 
 hence multiplying by i ; Bi and "e , and adding and dividing 
 a Y 
 by w’ we get 
 
 2 2 
 
 ‘ x 
 
 v 
 
 ——_—_— + s Ee 
 a’(w+ua) SP (wtup) xy (u+py) 
 which by substituting for w its value may be reduced to the 
 
 form 
 x - 3 
 SEL Ses Ve EP a- ., -- es SS ee Paes ol Es + ee ee 
 ataByn B+raByp ytaByu 
 which shews that the intersections of confocal surfaces of the 
 second order are lines of curvature, agreeably to Prob. VIII. 
 
 =1, 
 
 15. The curvature of the shortest line through any point 
 of an ellipsoid varies as the cube of the distance of the tangent 
 plane at that point from the center of the surface. 
 
 Since the osculating plane of the shortest line contains the 
 normal to the surface, the radius of curvature of the shortest 
 line will be the same as that of the normal section of the 
 surface which passes through its tangent, and therefore will 
 
 Woe Ge : 
 equal Soe ae rs PD=C by Prob. VII. p. 289. In the case 
 
 of a surface of revolution, the perpendicular on the tangent 
 
 plane coincides with the perpendicular on the tangent line to 
 
 the meridian, and consequently the radius of curvature of the 
 a?b? 
 
 meridian = —_, and the curvature of the shortest line is 
 
 proportional to the curvature of the meridian through the 
 
 same point. 
 
321 
 
 16. If a’, a’, be the parameters of the lines of curvature 
 that cross one another at any point of an ellipsoid (that is, the 
 semi-major-axes of the confocal hyperboloids on which they 
 are respectively situated) and @ and 4 7 — @, the angles which 
 the geodesic line through the same point makes with the lines 
 of curvature, then the value of a” sin®@ +a” cos’ @ is the 
 same for every point of the geodesic line. 
 
 It is shewn Prob. VIII. that the lines of curvature which 
 cross one another at any point of an ellipsoid are the curves in 
 which the ellipsoid is intersected by homofocal hyperboloids 
 of one, and of two sheets, respectively. If therefore we take 
 for the equation to the ellipsoid 
 
 i, 2 oe 
 at ow de ao: 
 we may consider the two lines of curvature at any point to be 
 
 determined by the hyperboloids whose equations are 
 
 we y 2 “ y? 
 
 “Sy eee g fone A 
 a” Pea? f—a” 
 
 = -a toa Le 
 gre BFL 2 cig 7—* 
 
 and a’, a’, are called the parameters of those lines of cur- 
 
 vature. If we subtract each of these latter equations from the 
 equation to the ellipsoid, we get 
 
 ue y” Poa 
 
 2, /2 (2 par (pe a, + 73 2) (,,/2 ae (2), 
 a”a (a -—b)(h -—a*) (a&-Cc)(a*-©&) 
 a y = 
 
 Es ata 6%) (62. 052) (abot) (Gia) © Hats) 
 But between the co-ordinates wv, y, x of a point in the ellipsoid 
 (1),.and 7° the square of a semi-axis of the diametral section 
 drawn parallel to the tangent plane at that point, we have by 
 Prob. XII. the relation 
 
 P x’ y rf 
 Gof) @-A@F-F) | @-A@oésr) 
 which compared with (2) and (3) gives for 7° the values 
 a’ —a’®, and a’ —a’™”. Now let P be the perpendicular from 
 
 _ the center on the tangent plane at ays, and 4, B, D the semi- 
 
 _ diameters of the diametral section, respectively parallel to the 
 lines of curvature and the geodesic line passing through zyz ; 
 then 
 
 21 
 a 
 
322 
 
 AB? ‘ par vir 2. | rapes 
 se = 4’ sin’ 0 + B’ cos’@ = (a — a”) sin’? 0 + (a — a”) cos’O 
 (ABPY? 
 
 or a’? — a” sin’ @ — a’” cos’*0 = (PD) = a constant, 
 since ABP is proportional to the volume of the ellipsoid, and 
 PD is constant by Prob. VII. p. 289. 
 
 The result a” sin? @ + a” cos’@ = constant, which is the 
 integral of the differential equation of the second order to the 
 geodesic line on an ellipsoid, was first given by Jacobi; and 
 the result PD = constant, as well as several other Problems 
 here given, are due to Joachimsthal (see Crelle’s Journal, 
 
 Vol. 26). 
 
 17. The tangents to the lines of curvature at any point 
 of an ellipsoid bisect the angles between the tangent lines to 
 the two circular sections through the same point. 
 
 For if the diametral section be drawn relative to that 
 point, the tangent lines to the circular sections are parallel to 
 the equal diameters of the section, and the tangents to the 
 lines of curvature to the axes of the section; and in an ellipse 
 the equal diameters are equally inclined to either axis. 
 
 18. To find the radii of curvature at any point of a 
 paraboloid. 
 
 Pr 2 
 Let the equation to the surface be — + = + 2%= 0; then 
 a 
 
 it may be shewn, as in Example 9, that the maximum and 
 minimum radii of curvature at a point wysx, are 
 
 va Sula genbisnee 
 
 " ape 2 
 a+b—2% y | 
 
 a GB Jab (S48 41). 
 
 Hence if the radii be bestia on the axis of x, the sum 
 of the projections = a + b — 22. 
 
 19. In surfaces of the second order, the two curvatures 
 are in the same direction at every point of the same surface, 
 or they are in opposite directions at every point. 
 
323 
 
 The curvatures are in the same or opposite directions 
 at any point of a surface, according as r¢ — s’ for that point 
 is positive or negative. But, substituting for 7, s, ¢ their 
 values from 
 
 2 1 
 ee ey we find rt—s? = X, ; ; 
 Camny 3.8 iry x apy 
 a quantity which can never change its sign by the variation of 
 2, y, x, and will always have the same sign as ary; therefore 
 for the ellipsoid and hyperboloid of two sheets, it will be 
 positive, or the two curvatures at every point of the surface 
 will be in the same direction ; and for the hyperboloid of one 
 sheet it will be negative, or the two curvatures at every point 
 will be in opposite directions, Similarly, for the elliptic and 
 hyperbolic paraboloids the curvatures are respectively in the 
 same, and opposite directions, at every point. 
 
 20. If # be the radius of absolute curvature at any 
 point of a curve defined by the intersection of two surfaces 
 u,=0, U,=0; and r,7,, be the radii of curvature of the 
 sections of w, = 0, w,.=0 made by the tangent planes to 
 U,=0, U,=0 respectively, at that point; it is required to 
 express R in terms of r,7,, and @ the angle between the 
 tangent planes. 
 
 Let OP, O,P (fig. 54 bis) be the normals at P to the two 
 surfaces, PV, PV, the radii of curvature of the sections of each 
 surface made by the tangent plane to the other; then if VO. 
 V,O, be perpendicular respectively to PV, PV,, by Meunier’s 
 Theorem O, O, will be the centers of curvature of the normal 
 sections; and if P# be perpendicular to OO,, R is the center 
 of curvature of their intersection; .«. PR=R PV=*r 
 PV,=r, OPO, =0, OPR = a suppose, O,PR = 6-a, 
 
 PVcosa rcosa cosa. sin@ 
 
 SLE RO 8 008, rear a ott EW R — ) 
 
 cos(@—a) sin@ cos@cosa_ sin@sina cos@sin@_ sin@sina 
 > = = qi —4+ ——_ = —_—___—_— 
 
 ite v; R R r POA 
 
1 1 2cos@ 1 
 
 SS = oo ke 7 
 
 Pr rv; r, 
 
 21. The locus of the focus of an ellipse rolling along 
 
 a straight line is a curve such that if it revolve about that line, 
 the sum of the curvatures of any two normal sections at right 
 angles to each other will be the same at every point of the 
 
 surface generated. 
 
 Suppose the ellipse to touch the line ON along which it 
 rolls in P; join SP and draw SN perpendicular to ON, then 
 SP is a normal to the locus of S, and if ON=2, NS =y, 
 
 ds 
 and the arc of the curve described by S =s, we have SP=y—; 
 
 faa tel Nie Ca aR or aiteknaan 
 WP SN” ds yy 
 eae cL aM Pai de ae Oe 
 ds°’ ds y’ r y 
 if r be the radius of curvature at S' to the locus of S; 
 ts 2a (25+) =2 08g, ee ae 
 Ses SP. %2-a 
 
 which shews, since SP and r are the two radii of curvature at 
 SS’ of the surface of revolution, that the sum of the curvatures 
 of any two normal sections at right angles to one another 
 is the same at every one of its points. 
 
 22. If # be the radius of curvature at a point whose 
 distance from ‘the axis =p, of the shortest line traced on 
 a surface of revolution; and if the radius of curvature of the 
 meridian at that point = r, and the part of the normal inter- 
 
 ‘ : 1 1 1 ] c 
 
 cepted between it and the axis =m, then — = — + e _ | = 
 
 ) Kier Tee 
 
 e being the value of p where the shortest line cuts the meridian 
 at right angles. 
 
 23. If a line of curvature be a plane curve, the tangent 
 planes applied to the surface at its various points will all make 
 the same angle with its plane. 
 
Jas! Necle teulp. 362 Strand . 
 
Hymeas Analytical Geometry 
 
 Jos! Neele teudp. 352 Strand . 
 
 a a 
 
Plate 3 
 
 Ja*Nede se. Burleigh Str. Strand 
 
Hymers Analy tical Geomeby 
 
 Plate 3 
 
 Ja? Neele se Burleigh Str, Strand 
 
> 
 
 rar 
 
 bs 
 S 
 ne 
 Sd 
 E 
 3 
 5 
 % 
 = 
 S! 
 8 
 S 
 
Flate 4. 
 
 ale sc. Burleiah Sta: Sand 
 
 JatNve 
 
Plate &. 
 
 Jos Neele teulp. 862 Strand. 
 
Liymers Analytical Geometry 
 
 Plate 4. 
 
 56 bis 
 
 Jost Neele teulp. 862 Strand. 
 
& 
 3 
 & 
 J] 
 f 
 ¥ 
 
Hymers Analytical Geomary r 
 
 y See Jost Neele fiulp, 852 Strand. 
 
 = Sp wee 
 
Cambridge. 
 
 LIST OF BOOKS 
 
 PUBLISHED BY 
 
 MACMILLAN AND CO. 
 
 Axspy on Roman Civil Procedure 
 Aischyli Eumenides. Drake. 8vo. cloth . 
 Aristotle on the Vital Principle. Collier. Cn. 8vo. cl. . 
 
 Breamont’s Story of Catherine. Fep. cloth 
 Birks’ Difficulties of Belief. Crown 8vo. cloth . 
 Bolton’s Evidences of Christianity 
 
 Boole’s Philosophy of Logic. Preparing 
 
 Differential Equations. Preparing ‘ 
 
 Mathematical Analysis of Logic, sewed . 
 
 Brave Words for Brave Soldiers and Sailors, per 100 
 Butler’s (Archer) Sermons. Ist Series, 8vo. cloth 
 —— 2d Series, 8vo. cloth 
 
 —— Lectures on Philosophy. 2 vols. 8vo. cl. . 
 
 Letters on Romanism, 8vo. cloth 
 
 CAMBRIDGE Theological Papers. Moor. 8vo. cloth. 
 Cambridge Senate-House’ Problems and Riders, with 
 Solutions :— 
 1848 to 1851.—Problems. Ferrers and Jackson 
 1848 to 1851.—Riders. Jameson 
 
 S&S 2-6 ts 
 
 OS Bri oa eS 
 
 Oo © 
 
 i 
 SP IS 
 
 ont pp 
 
 2 B&B @& 
 
 So kG (OS Oe 
 
Zz LIST OF BOOKS PUBLISHED BY 
 
 Cambridge Senate-House Propione and Riders, with 
 Solutions :— 
 
 1854.—Problems and Riders. Walton and Mackenzie 
 
 1857.—Problems and Riders. Campion and Walton 
 
 Cambridge Fitzwilliam Museum Hand-book, en. 8vo. sd. 
 
 Cambridge Mathematical Journal. Vol. L., 8vo. cloth. 
 
 Cambridge and Dublin Mathematical Journal. 9 vols. 
 
 Campbell on the Atonement. 8vo. cloth 
 
 Cicero on Old Age, Translated. 12mo. sd. 
 
 Cicero on Friendship, Translated. 12mo. sd. 
 
 Colenso’s Journal in Natal. Fep. cloth 
 
 — Village Sermons. Fep. eluth . 
 
 Ordination Sermons. 18mo. sd. 
 
 Companion to Communion. 18mo. cloth 
 — morocco 
 
 ——: limp cloth 
 
 Cooper’s Geometrical Conic Sections. Cn. 8vo. cl. . 
 Cotton’s Sermons at Marlborough. Fep. cloth 
 
 Crosse’s Analysis of Paley’s Evidences. 18mo. . 
 
 Davies on St. Paul and Modern Thought. 8vo. 
 Demosthenes de Corona. Drake. Cn. 8vo. cl. 
 —— Translated by Norris. Cn. 8vo. cloth 
 
 Drake’s Notes on Jonah and Hosea. 8vo. cloth 
 
 Drew’s Geometrical Conic Sections. Cn. 8vo. cl. 
 Evans’ Sonnets on Wellington, 8vo. sd. 
 
 FARRAR on Classical Studies. 8vo. sd. 
 
 2 & 
 0 10 
 Oies 
 Ove 
 0 18 
 7 -4 
 0 10 
 0 2 
 0.2 
 0 5 
 ee 
 OPst 
 Orage 
 Was 
 Use t 
 er) 
 Oe? 
 0. 5 
 Os 
 aig 
 0 4 
 ee 
 Ort 
 
 Os Ge Ge Gs Cy 80 Fes FCs OI OO Gs, Ca Ce 
 
 Con 
 
 ery SS} Nees aes =n 
 
 SS, ee 
 
MACMILLAN AND CO. 
 
 Farrar on Truth making Free, sd. . : 
 
 Flowers of the Forest. Cn. 8vo. sd. 
 
 Frost’s First Three Sections of Newton. Cn. 8vo. 
 
 GopFray on the Lunar Theory. 8vo. cloth 
 
 Grant’s Plane Astronomy. 8vo. 
 
 HALE on Transportation. Crown 8vo. sd. 
 
 Hallifax’s Analysis of Civil Law. Geldart. 8vo. 
 
 Hamilton on Truth and Error. Crown 8vo. cloth 
 
 Hare’s (Archdeacon) Charges. 3 vols. 8vo. cloth . 
 
 —— Charges for 1843-45-46.: 8vo. cloth 
 —— Miscellaneous Pamphlets. 8vo. cloth 
 
 Victory of Faith. 8vo. cloth . 
 
 — Mission of the Comforter. 8vo. cloth 
 — Vindication of Luther. 8vo. cloth . 
 Contest with Rome. 8vo. cloth - 
 
 Parish Sermons. 2d Series. 8vo. cloth 
 
 Portions of the Psalms. 18mo. cloth 
 
 Hardwick’s Middle Age Church History. Crown 8vo. 
 
 History of the Reformation. Crown 8vo. 
 —— Twenty Sermons. Crown 8vo. cloth 
 
 en Christ and Other Masters :— 
 
 Part I. Jntroduction. Part Il. The Religeons 
 of India. Part IIT. The Religions of China. 
 
 8vo. cloth. each 
 
 Hemmine’s Differential and Integral Calculus. 8vo.. 
 
 Hervey on the Genealogies of our Lord. 8vo. cloth . 
 
 Sermons on Inspiration. 8vo. cloth 
 
 on University Lecturers, 8vo. sd. 
 
 a Kae 
 
 Se an HS SF CO SS LOS eee OS et i OS 
 
 ao aS aS 
 
 10 
 
 Or 
 
 io) 
 
 10 
 
 aa Co 
 
 >) 
 
 Oo 
 
 OS-O A 2) Bm - SO 6) OO OF eer cS 
 
 ‘ane (2p eth Ne ee 
 
4 LIST OF BOOKS PUBLISHED BY 
 
 Howard on the Book of Genesis. Crown 8vo. cloth . 
 
 Exodus and Leviticus. Crown 8vo. 
 
 — Numbers and Deuteronomy. Crown 8vo. 
 Howes’ Chureh History of First Six Centuries . 
 
 Humphreys’ Exercitationes Iambice. Fep. cloth 
 IneteBy’s Outlines of Logic. Fep. eloth . 
 
 JuWELL’S Apology. Translated by Russell. Fep. cd. 
 Journal of Classical and Sacred Philology, Vols. I to 
 Lites . , . , each 
 Three Numbers. Published yearly, 4s. each. 
 Justini Martyris Apologia Prima. Trollope. 8vo. . 
 Juvenal, with Notes, by Mayor. Crown 8vo. cloth . 
 
 KENNEDY on International Law. Crown 8vo. cloth . 
 Kingsley’s Two Years Ago. 3 Vols. crown 8vo. cloth 
 —— Westward Ho! Crown 8vo. cloth 
 
 —— The Heroes. Crown 8vo. cloth, gilt leaves 
 
 —— Glaucus. Fep. cloth, gilt leaves 
 
 —— Alexandria. Crown 8vo. cloth 
 
 —— Phaethon. Crown 8vo. . 
 
 Law on the Fable of the Bees. Maurice. Fep. 
 Lectures to Ladies on Practical Subjects. Crown 8vo. 
 
 Lushington’s Poems. Fep. 8vo. cloth 
 
 Mackenzie on the Influence of the Clergy 
 
 Maclear on Incentives to Virtue 
 
 2 se 
 0 8 6 
 010 6G 
 0 5 6 
 0 3 6 
 Oy nt ou 
 012 6 
 0 7 
 010 6 
 URNA es 9 
 Le Live 
 Or) Zeneg 
 0°76 
 > os 
 0 5 0 
 O- 2-90 
 0 4 6 
 Oa 
 
 3 0 
 O38, Goma 
 Quy Leah 
 
MACMILLAN AND CO. 
 
 Mansfield’s Paraguay, Brazil, and the Plate 
 Mansfield on the Constitution of Salts. Preparing. 
 M‘Coy’s Contributions to Paleontology. 8vo. cloth . 
 Masson’s Essays on the English Poets. 8vo. cloth 
 MauRIcE ON THE OLD TESTAMENT :-— 
 The Patriarchs ana Lawgivers. Crown 8vo. cl. 
 The Prophets and Kings. Crown 8vo. cloth 
 —— on THE New TESTAMENT :— 
 Gospels of St. Matthew, Mark, and Luke, and 
 the Epistles of St. Paul, Peter, James and 
 Jude. 8vo, cloth ‘ 
 Gospel of St. John. Crown 8vo. cloth 
 Epistles of St. John. Crown 8vo. cloth 
 —— Theological Essays. Crown 8vo. cloth 
 —— The Prayer-Book. Fep. cloth 
 —— The Church a Family. Fep. cloth 
 —— The Lord’s Prayer. Fep. 8vo. cloth 
 —— The Sabbath Day. Fep. 8vo. cloth 
 —— Christmas Day, and other Sermons. 8vo. cloth 
 —— Doctrine of Sacrifice. Crown 8vo. cloth . 
 —— Religions of the World. Fep. cloth 
 — Kcclesiastical History. 8vo. cloth. . 
 —— Learning and Working. Crown 8vo. cloth 
 —— A Photograph Portrait . 
 Mayor’s Lives of Ferrar. Fep. cloth 
 
 Autobiography of Robinson. Fep. cloth . 
 
 Juvenal. Crown 8vo. cloth 
 
 Minucius Felix, translated by Lord Hailes. Fep. cl. 
 
 oo Oe On On ONO £O8° Sono gO § oS SO 
 
 10 
 
 10 
 
 Soe Coe Cle Co pec BeO> BC: GECo gC>: FS 
 
 -_-~ 
 we 
 
6 LIST OF BOOKS PUBLISHED BY 
 
 Napikr on Bacon and Raleigh. Crown 8vo. cloth 
 
 Nind’s Sonnets of Cambridge Life. 
 
 —— Odes of Klopstock. Fep. cloth 
 
 Norris’ Ten School Room Addresses. 
 
 Norway and Sweden, Long Vacation 
 
 Parkinson's Elementary Mechanics. 
 
 Crown 8vo. cloth 
 
 German Lyrist. Crown 8vo. cloth . 
 
 Letters from Italy and Vienna. Fep. cloth 
 
 18mo. sewed . 
 
 Ramble in 
 
 Crown 8vo. cl. 
 
 —— Elementary Optics. Crown 8vo. cloth 
 
 Parminter’s English Grammar. Fecp 
 
 Payn’s Poems. Fep. cloth 
 
 . cloth 
 
 Payne’s Decaeus and The Bond Child. 18mo, cloth . 
 
 Peace in War. Crown 8vo. sewed 
 
 Pearson’s Finite Differences, 8vo. 
 
 Perowne’s Elementary Arabic Grammar. §8vo. cloth. 
 
 Perry’s University Sermons. Crown 
 
 8vo. cloth 
 
 Phear’s Elementary Mechanics. 8vo. cloth 
 
 —— Elementary Hydrostatics. Crown 8vo. cloth 
 
 Plain Rules on Registration. Crown 
 
 8vo. Per 100 . 
 
 Plato’s Republic. Translated by Davies and Vaughan. 
 
 Principles of Ethics. Crown 8vo. sewed . 
 
 Procter on the Prayer-Book. Crown 8vo. cloth 
 
 Puckle’s Conic Sections. Crown 8vo. cloth 
 
 Ramsay’s Catechiser’s Manual. 18m 
 
 o. cloth 
 
 Reichel’s Sermons on the Lord’s Prayer. Crown 8vo, 
 
 Robinson on Missions and the State. 
 
 SaLLust, with Notes, by Merivale. 
 
 Fep. 8vo. cloth. 
 
 Crown 8vo. cloth 
 
 ram at AN = EY om Re 
 
 ne RE de a or pe ee ee aes aes Eee 
 
 0 
 
 0 
 0 
 0 
 
 0 
 
 A ognmnnrnrocdcoen ss 
 
 key er ep) Mesh eee) ieee tee fee 
 
 =) 
 
MACMILLAN AND CO. 
 
 Selwyn’s Work of Christ in the World. Cn. 8vo. sd. 
 Verbal Analysis of the Bible 
 
 Simpson’s Epitome of Church History. Fep. cloth 
 Smith’s (Alexander) City Poems. Fep. cloth. . 
 Smith’s (Barnard) Arithmetic and Algebra. Cn. 8vo, 
 
 Arithmetic for Schools. Crown 8vo. 
 
 . 
 
 Key to Arithmetic. Crown 8vo. 
 
 Snowball’s Trigonometry. Crown 8vo. cloth. 
 
 Introduction to Trigonometry. 8vo. sewed 
 
 Cambridge Course of Natural Philosophy 
 Swainson’s Hand-Book to Butler’s Analogy. Cn. 8vo. 
 
 Tait and STEELE’s Dynamics. Crown 8vo. cloth. 
 Taylor's Restoration of Belief. Crown 8vo. cloth 
 
 THEOLOGICAL MANUALS :-— 
 
 Hardwick’s Church History of the Middle Ages 
 
 The Reformation ; A : J 
 Howes’ Church History of the First Six Centuries . 
 Procter on the Prayer-Book 
 Westcott on the New Testament Canon 
 
 Thring’s Construing Book. Fep. 8vo. cloth 
 
 —— Elements of Grammar. 18mo. cloth 
 
 —— Child’s Grammar. 18mo. cloth 
 
 Thrupp’s Psalms and Hymns. 18mo. cloth 
 
 limp cloth 
 
 Antient Jerusalem. 8vo. cloth 
 Thucydides, Book VI. Frost. 8vo. cloth 
 Todhunter’s Differential Calculus. Crown 8vo. cloth . 
 
 —— Integral Calculus. Crown 8vo, cloth 
 
 So cofo. 6 S&S SFC OBA Se 
 
 —) 
 
 = 
 
 os oe BS ee  O 2 OLS ES 
 
 laa az H 
 Ci 4 = St Sr bee EO 
 
 10 
 12 
 
 = bS — bo 
 
 15 
 
 “I 
 
 10 
 10 
 
 for) 
 
 Sse sy Oe ee OS i Ga C2: 9 
 
8 BOOKS PUBLISHED BY MACMILLAN AND CO. 
 
 Todhunter’s Analytical Statics. Crown 8vo. cloth 
 —— Conic Sections. Crown 8vo. cloth . 
 
 —— Algebra. Crown 8vo. 
 
 Tom Brown’s School Days. Crown 8vo, cloth . 
 Trench’s Synonyms of the New Testament. Fep. cl. 
 —— Hulsean Lectures. Fep. cloth 
 
 —— University Sermons. Fep. 8vo. cloth 
 Vaughan’s Parish Sermons. Crown 8vo. cloth . 
 
 Waters of Comfort. Fep. cloth 
 Westcott on the New Testament Canon. Crown 8vo. 
 
 —— Introduction to the Study of the Gospels. 
 
 Westcott and Hort’s Greek Testament. Crown 8vo. cl. 
 
 Wilson’s Five Gateways of Knowledge. Fep. cloth 
 —— People’s Edition, ornamental stiff’ covers 
 Wilson’s Dynamics. 8vo. boards 
 
 Wright’s Hellenica. Fep. 8vo. cloth 
 
 —— Help to Latin Grammar. Crown 8vo. cloth 
 —— Seven Kings of Rome. Fep. cloth . 
 
 —— Vocabulary and Exercises. Fep. cloth 
 
 Rk. CLAY,’ PRINTER, BREAD BTREET HILL, LONDON, 
 
 £ s. 
 0 10 
 0 10 
 0 10 
 0 5 
 0.5 
 0 2 
 0.5 
 0 4 
 012 
 0 2 
 0 1 
 0 9 
 0 3 
 Ogee 
 0 3 
 0 2 
 
 a QS Ba GB 
 
 ep Sy ery eS ep SOS 
 
; : . 
 ( ie | 
 a iM | 
 ¢ 1 | 
 , me 
 : iv i 
 
Hovis y 
 Rae ag 
 
 7” 
 te WN eee . 
 if m He ai 4 
 fe} Ni SEN A) mn 
 ‘he ig 5 ty oe i" pe eee i 
 , rar ul UAE eee ia ay 
 
 i 
 fog see 
 
 ees 
 f fia AL 
 
i 
 VE 
 
 i | . ’ 76 
 a eat an 
 on ity ei ( 
 Cui 
 an 
 
 een 
 
 es 
 
i XY 
 { ‘ 
 ie e,/ 
 behet 
 ve ur 
 .*) j ‘ 
 wy : wegen | i oe f Pay ; ; ha , Y j pi ( my bt fu 
 isi te ee teed ed . vie ( . ; ] i! 4 \ 
 OMAR HACERA LHCD RR Senne §=—516.33H99T1848 Dra ete este sOA bast 
 bier f Weer richeice deurnsedens yur leuskedsbeizdetsfybrieheats A TREATISE ON ANALYTIC rela Be Al 
 My “ah ert shy ystiin Serra be Ca any ( : mhatiatts int “ir ADA tiles tah 
 ite he pee mts Beir asnt an ; been chet se 
 Fatah Fur hs 1 si ithe fh Ge iQ tied by kill? hime ds ( ier 
 dda a ere a Ah rhs ieaes Carpe ced 
 ‘7 DP ¥ M4 Se Pas " \ i t Ay j ( ’ 
 vel Byun PAM vi 3 0112 Hy ' 
 seit rk H i 
 ‘ Ws Ps ane ue j 
 hpi rue 
 i fehl tebe tl 
 i wide 
 bet Artbed 
 hehe A Ye i 
 { ae et 
 ea Ry Ry 
 , Girt t 
 Lf pe Le ayy! 
 selieheie ts he | { A 
 e WV {alts Wien | Weld 
 i a HURL! is 
 Pedegets i rs : 
 A) ntyrye I vet vet 
 et" L Te ee Wee 
 htt Liha) ! iu Cee 
 Pe aba depp 
 \ yeh t 
 
 Vey 
 Mish uP i 
 
 the |} ¥4 i ah bes 
 
 4 “Witch he re aN $ 
 A vayiaiist iitiedie es bedeliit j PN hed 
 Vasant a ect rete y Hy purl a 
 
 Alb 
 ie aa 
 Nea alte 
 A 
 Joie tein bebe t 
 pee “yet ie 
 betvievelt reed Hs 
 tte wold ed! fe 
 : \ He jieily tir 
 beeeged ) 
 pave ea re a fe Gen 
 Wig eds ihe vi 
 
 ure ii! iva 
 bate been ts (Fie 
 
 i] 
 beet ff ' 
 Hee WEE MOIS Ie Be thelial 
 Hi(PERE LED Mee pad ei 
 a det Wane die het 
 
 iebtetié } 
 + petals te Pete dete 
 eat Meus : 
 Petcnht Welt ey 
 
 hat 1 
 th “ rsies ap ye 
 hide 18 Abele tava 
 ‘ ih} : 
 
 Dies aa et 
 
 pyle it Payee 
 ; eitocgeweeshinht ptaehet 
 Peta pape leery ‘ b ybely pile 
 : ir a BEAT AL Pe be He tee yet 
 eee 
 pM i 
 Wh : 
 ny untae fl 
 Aga) ‘eh 
 
 beh shit 
 
 sin 
 
 Puce 1 Wetint) 
 Qepomedt- tev Milk 
 We Ply heey Deg 
 we el 
 th iY 
 He edith eyeye eee ben 
 Watied ent! writs ete i i t 
 Vata te We Ah { 
 
 “line 
 it (ih ih Neh ¥ 
 “Ele Maly MOM 
 At {) Heaps eis 
 
 nah Ge thet hath | 
 Tea ie thle Thetis 
 Pi debe hehe: 
 
 \| Ped tp heel i a lt Dont * 
 PA a ot nae oO A CU vary Ny §) iite ped “et 
 ie tlefi POMS MER de ftied eben eds eH et 
 
 Ale edie She tie WAGER Be fle fie Gs atti 1A 
 iat Rohertety Age we Avett 
 rPfediay ths te i AP A” i 
 
 ‘ Wee ites 
 eh Seiten | 
 
 jenn Pele tye de He pivede Wye 
 Nr ba We Wed { eHEM i Al 
 rY Uy Fabs (et Ve tha ths i [ VE tlette ry ete et 
 y HY fehited ty ¢ " 
 
 HM ia ATTA Thy Sha eae UME fivk 
 
 fy DHA AS ee Di left Bete & 
 
 Hetbegi-He\ Hie Heyy } 
 ‘ Mi tesa dene he 
 Mie Ginliepe ont et $ vninanany fi ( Web 64 aie 
 Wey Media teMe beg ie {it a ae Me tvegir He a he hetisleedbe Getto g i aanies (st 
 hehe “AP edi fb ied tiene ‘ ’ : ee Reel nw 
 Depeliebed Pein U hia janice eee be i vele ie (tore! Hi 
 
 ii hea fhe iinfarte (ee edger 4, 
 sete pe tehh UA 
 Teese ise 
 
 Ke de ted cdiotis teil de he 
 i) At i ana { 
 ted WAN Me Be fi ite Ube 
 
 Hittite he abcd Hef 
 
 Dp ed ened Osher eeGee 
 
 Wey ht Ht 
 ines etitets Sth Meat fled epee hey 
 ' i te lis \ a im fief tt eh ‘\ rit 
 ye BiH} Wen M (A ¥ 
 Aye a 
 
 Gj 
 
 Uy jotpeiey! M4 { 
 
 vat uata| ibe hit ne an 
 d u t 
 
 Wedd dy eye preted! se fet tye fled #fls fi a aie fi hh 
 Ph helatey Hal Py asia) iy AA " Ahem ah Aes ote 
 Hehe ee i HO ik rete t f aa ee ia. hi Lak jietiaet Ness (EAE Ie Yb th 
 Atte hehe wer tif itp fe [Cine edicte We ke W ey) A BeVe db Py hd 
 GMD EYER HED ete 8A) Ue oh) et) Rv fe feeriedey { es hed Hi ) at 
 Gira iehes Pe heb see Mog ay iby Me be tele eget cok ied Ha eon We, 
 be vate Ce oan} rin aba it Wo mbral the abet Yad é ; } Nigh ‘ 
 HN fre ple Ssh hee Pye ef HN Pelle het F 
 Wd [Eble uy 4 To aawet é Hi ad Pea len ee BY Sebel Woh 
 ble peseqe Ri td ehh ica ith Poien its f| yA Ae3 dani Hehofated, CRA Tes Oe W i 
 } Jit hie am tia iy hen geal) penesie he has Vest aay Teen TL MR Vilernok | 
 shy) l Pare eh ee ee ' J Heh fon fee h Bopha | 
 wi gOS eae i wie ii jiates Hi¥) iM bt ta geiinde tout ae ui tone et } CSW Ay APR ALE Sd ah 
 
 he Ge Babine ate th iy) 
 
 (Oo tb. Beach eh (ted 
 rt) 4 Nia! rae Heterdle Le 
 tt uJ 1a 
 
 t Hetiel ah 
 
 f yay Nitra en | 
 rfletlepede te hepatic i ht Cater A HE Le bebed hehe tebe poflbbie de” 
 [eiRecwell pete the ieee the woh gpl ag 
 
 (frre tegit Ue Pee AE Oiletio’ 
 
 ba Dep ete 
 J 6 
 
 ‘pee {Sit 
 
 De hepewe sacl a a Wee tied bebe eet 
 ( it ito isi eb wt f fet eh He ed tered etl iets aly ' 
 +f i net ei (the i 
 yi) Heh! ei? it 
 vie ahh ib aie ted Cid es ead (a 
 a 6 Pee uid Hh 
 help eehte ahr he Neti he ijell Hopepbataet it 
 dhe Rieke te be ied pricy She Getuned: thy tt 
 ( Mifieht ie 1 a i 
 
 PAL 
 
 He jlr De t 
 he pity eae f i} 
 feditien ata ie 
 eWele Hedin fesy Hehe hed if ave 
 ¢ at HEME AMY edb i i 
 GA LENS eA beep Levee a! oe 
 LEASE elfemd may eM at af 
 j as 
 et 
 
 wt 
 v4 
 
 phe hag ibe Ooh 
 Hopieedel Wed ott de tofle } 
 ; i ¢ 4 
 Vege eu ] 4 dee ; 
 
 on if, rai Pitan We { ; 
 
 Jrebale bes Meni Aa oe ary 4 
 
 ie ava hee BORE WON Fgh 
 
 TP Poa payed 
 
 oy