'^nOMINGTON 
 
 §P^ r M\ 
 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 s 
 
 
 
 
 
 
 
 
 
 . 
 
 
 
 
 
 
 
 
 
 
 
 
MERCURIAL 
 
 AIR-PUMP. 
 
 SHAWM'S p. THOMPSON. D.SC.,B.A., 
 
 PRINCIPAL OF, AND PROFESSOR OF PHYSICS IN, THE CITY AND GUILDS OF LONDON TECHNICAL COLLEGE, FINSBURY. 
 
 E. & F. N. SPON, 125, STRAND, LONDON. 
 
 NEW YORK: 35, MURRAY STREET. 
 
 1888. 
 

 
 
 
 
 
 
 
 
 
 
THE DEVELOPMENT 01 THE MERCURIAL AIR-PUMP. 
 
 j. y o o 
 
 :6^\ejs i aM 
 
 Within the range of the physical sciences 
 there are many subjects of interest to the 
 investigator, some of them of importance in 
 the technical industries, which yet are little 
 known, save to the close student of scientific 
 literature. An idea occurs to some original 
 worker, who makes experiments, records 
 results, draws inferences, and embodies his 
 investigations in a memoir or note in the more 
 or less obscure journal of proceedings of some 
 scientific society, where it remains buried amid 
 a heterogeneous mass of like matter. A few 
 years later another worker, not knowing what 
 has been done, stumbles across a similar idea, 
 constructs his apparatus, makes his experi¬ 
 ments and observations, and in turn consigns 
 his work to a similar temporary oblivion. 
 Years after, or it may be centuries, th 
 of science in other departmental- 
 point where it touches the unreq/embered work 
 of those who are gone. Were that -pork 
 available, were even a plam cluei-ti Ilt£\ 
 existence known, it would be aVpnce brought 
 into line and made useful. For >&nt q .>£ such 
 a clue it remains buried, and the 
 wait for some other to re-investigate ^ 
 already was a portion of acquired knowledge ; 
 and the useful application to industrial 
 purposes is delayed—perhaps for years. He 
 who will explore the dusty corners of science, 
 will hunt up the hidden treasures, arrange 
 them, co-ordinate them with more recent know¬ 
 ledge, and render access to them easy for the 
 busy workers around him, performs a service 
 which, though it brings no renown, is at least 
 useful.* In such dusty corners of science lie 
 the scattered fragments of the literature of the 
 mercurial air-pump. 
 
 All true mercurial air-pumps are, of course, 
 based upon the principle of the barometer; 
 that is to say, in all of them the vacuum is 
 constituted in the manner invented by Torri¬ 
 celli, namely, in an enclosed space above a 
 barometric column. In all of them the object 
 is to render the Torricellian vacuum as perfect 
 as possible. In some of them the object is 
 effected by driving the air upwards out of the 
 barometer head by raising the barometric 
 column ; in others, the air is forced downwards 
 by the injection of more mercury into the 
 
 * The author here desires to express his indebtedness to 
 the many friends who have assisted in collecting this informa¬ 
 tion ; also to the valuable paper by Herr E. Bessel-Hagen, 
 referred to in the text. He also desires to refer to a memoir 
 by Professor Hellmann, of Riga, a copy of which he received 
 pnly after the greater part of this paper was already compiled, 
 
 barometer-head. In others again, the air is 
 pushed up one barometric column and down 
 another. In some recent kinds of mercurial 
 pump, several of these forms are combined. 
 Other pumps again depend on the injection 
 of mercury at high pressure through an orifice. 
 These distinctions will furnish us with a basis 
 for classification. Before attempting this, 
 however, a few general historical notes may 
 be given. 
 
 In 1643, Torricelli discovered the possibility 
 of producing a vacuum above the top of a 
 mercury column, by filling with mercury a tube 
 closed at one end and then inverting it into a 
 cup containing mercury. Until 1650—a period 
 of seven years—when the mechanical air-pump 
 was invented by von Guericke, there was no 
 way of producing a vacuum, and the 
 loreptibe Academicians, who made in the 
 internal mWiy experiments on the properties 
 of vacuous »pace, employed as their pump a 
 1 ^rp^rijcellian tube (Fig. 1) enlarged at its upper 
 
 Fig. 1. 
 
 QlS* ^ 
 
 A 
 
 B 
 
 Tube Used by Florentine Academicians 
 for Exhausting. 
 
 part into a bulb or reservoir closed at the 
 summit by a cover, luted on with some suitable 
 sort of cement. The apparatus is figured in 
 Daguin’s “ Traite de Physique,” vol. 1, p. 278, 
 (edition of 1855). Writing in 1855, Daguin 
 states that though the employment of the 
 vacuum obtained by the lowering of a mercury 
 column had been perfected by Baader and by 
 Hindenburg, their machines required such a 
 large quantity of mercury that they had been 
 abandoned in favour of the mechanical air- 
 
2 
 
 pump. The Torricellian method of exhaustion 
 was pursued however, whenever a very perfect 
 vacuum was required, for example, by Count 
 Rumford,* in his researches upon the propa¬ 
 gation of heat. The bulb to be exhausted he 
 sealed to the top of a barometric tube having a 
 constriction near the top ; it was then filled 
 with hot recently-boiled mercury and inverted, 
 when the mercury left the bulb vacuous. It 
 was then sealed off before the blow-pipe just 
 as exhausted glow-lamps are sealed off to-day. 
 An arrangement almost identical with that of 
 the Florentine Academicians was suggested 
 as a novelty so late as 1810, by Traill.f In 
 1845, Edward A. KingJ patented the con¬ 
 struction of an incandescent electric lamp, 
 having a carbon pencil situated in the 
 Torricellian vacuum at the upper end of a 
 barometer tube, a further suggestion being 
 made for sealing the exhausted lamp. Amongst 
 still more recent methods of producing in a 
 glow-lamp a Torricellian vacuum without the 
 aid of any specific pump, maybe noted that of 
 Andre,§ who seals the lamp globe to the top of 
 a barometric tube, another short tube opening 
 into the bulb being sealed on above. Mercury 
 is driven from below until it fills the lamp, 
 when the upper tube is temporarily stopped, 
 the mercury being then lowered, creating a 
 vacuum into which the occluded gases are 
 liberated; the mercury is then raised, and these 
 gases expelled through the stopper, and the 
 lamp is sealed at the top ; then the mercury is 
 lowered and the lamp is sealed off at the 
 bottom. Swinburne|| has also tried this method, 
 but has not found it successful. None of these 
 processes of utilising the Torricellian vacuum 
 amounts to the invention of a distinct Torri¬ 
 cellian air-pump. Although most of the 
 mercurial air-pumps to be described started 
 from the fundamental notion of producing a 
 barometric vacuum, it seems possible, if not 
 probable, that some, at least, of the various 
 forms had a different origin. Two of the main 
 defects of the ordinary mechanical air-pump, 
 are the faulty packing of the pistons and the 
 unavoidable clearance-space between the 
 piston and the end of its cylinder. It naturally 
 occurred to several experimenters, possibly to 
 many, that both these defects might be obviated 
 by covering the piston with a layer of mercury. 
 Air-pumps thus provided with a mercury 
 
 * Rumford. Essays, vol. ii., p. 393. (Edition 1798.) 
 
 + Traill. “Nicholson’s Journal,” xxi., pp. 63 and 161. 
 
 t King. Specification of Patent, No. 10919, 1845. 
 
 ? Andre. Specification of Patent, No. 4654, 1881. Electric 
 Illumination, vol. i., p. 665. 
 
 II See Electrician , xix., 159, July, 1887, 
 
 packing have been described by Kravogl * for 
 exhausting, and by Regnault t and Cailletet 
 for compressing. In the following sketch, 
 however, attention is confined to the barometric 
 species of pump. It will be convenient to 
 classify these machines under the following 
 heads:— 
 
 Classification of Mercurial Air-pumps. 
 
 I. Those which drive the air up a barometric 
 tube. 
 
 II. Those which drive the air down a 
 barometric tube. 
 
 III. Those which drive the air up one 
 barometric tube and down another. 
 
 IV. Combination pumps. 
 
 V. Injection pumps, dependent in their 
 action upon the velocity of efflux of a stream 
 of mercury. 
 
 VI. Mechanical mercurial pumps. 
 
 Under the first three of these classes there 
 
 are sub-classes comprising “ shortened ” 
 pumps. 
 
 Class I.— Upward Driving Pumps. 
 The oldest of all mercurial air-pumps belongs 
 to the class in which the air is forced upwards 
 above the top of a barometric column. It was 
 
 Fig. 2. 
 
 clISU o ^ **,«*** -- xr—j \ '' 
 
 Similarly a mercury packing was employed in a water-pump 
 by Haskins, and is described in Desagulier’s Course of 
 Experimental Philosophy (1763), vol. ii., p. 491. 
 f Regnault. “Relation des Experiences,” ii., 553 - 
 4 Swedenborg. “ Miscellanea observata circa res nati;- 
 rales.” Lipsise. 1722. Fig. 11. 
 
3 
 
 invented by Emanuel Swedenborg, the famous 
 theosophist, and is described in his “ Mis¬ 
 cellanea,”* published in 1722. The apparatus 
 and a drawing of the machine is given, which 
 is reproduced in Fig. 2. A glass receiver, B, 
 stands upon a plate, A, as in the ordinary 
 mechanical air-pump, and the plate, A, is 
 supported on three legs forming a tripod table. 
 ” Beneath this table,” to follow Swedenborg’s 
 own words, “ there is a certain conical vessel, 
 E, of iron, hollow, and accurately fixed on the 
 under side of the table so that it includes two 
 apertures, c and d, furnished with valves.” 
 The valve c, opens downwards, the valve d, 
 upwards. The lower end of the iron vessel 
 ends in a flexible leather tube,y^, to which is 
 adapted a piece of iron tube, gg, ending in an 
 open mouth piece, m. This iron tube was to be 
 alternately raised and lowered precisely as is 
 done in the modern Geissler or Toepler pump 
 with the mercury vessel. Swedenborg’s in¬ 
 structions are precise. 
 
 “ Operatio. —Immitte mercurium vivum per (m), 
 ilia copia, ut repleat (f) (f) et aliqualem partem E ; 
 si dein sursum levas (g), tunc mercurius ascendit in 
 E usque ad mensulam: si dein demittis idem (g) 
 infra altitudinem 28 pollicum, tunc descendit mer¬ 
 curius in E, attrahit secum aerem ex recipiente per 
 valvulam (d), et sic levando et demittendo, dum 
 omnis aer sit exantlatus. Habeas etiam foramen 
 quoddam in (2) sub mensula, quod aperias ope 
 trochleae, et immittas aerem cum velis.” 
 
 Further instructions say that the lifting must 
 be continued until some drops of mercury 
 come out at d. The nature of the valves c and 
 d is not stated; they appear to have been such 
 as were used in the current form of mechanical 
 air-pump. 
 
 In the above description and figure, Sweden¬ 
 borg’s lettering is adhered to, but it may be 
 convenient to state that in all the following 
 cuts the lettering adopted is as follows:— 
 
 A. — The “ pump-head ” or hollow globe, which is 
 
 filled and emptied. 
 
 B. — The barometric column or tube, also called the 
 
 “shaft ” of the pump. 
 b. —Barometric gauge. 
 
 D. —Drying apparatus. 
 
 E. —Eject-tube or valve. 
 
 f. —“ Fall-tube.” 
 
 G. —Gauge. 
 
 g . —The flexible lifting tube connecting s and b. 
 
 h. —Overhead-tube. 
 
 J.—Mercury valve. 
 
 * Swedenborg'. “ Miscellanea observata circa res naturales, 
 el praesertira circa niineralia, ignem, et montium strata.” 
 Lipsiie, 1722, p, 101, See also “ Gren’s Journal,” vol. iv., 
 
 p. 407. 
 
 K.—Collecting chamber. 
 
 l.— Pressure chamber at bottom of shaft. 
 
 M.— Second chamber for partial vacuum. 
 
 n.— “ Side-tube.” 
 
 p.—Piston. 
 
 p .—Plug of rubber. 
 
 r.— “ Receiver.” 
 
 r. —Regulating pinch-cock. 
 
 s. —“ Supply-vessel,” or reservoir of mercury, 
 j.—Rubber packing. 
 
 T.— Tap. 
 
 t. —Air-trap. 
 
 u. —The valve opening inwards from the exhaust 
 
 tube to the lower end of the pump-head. 
 
 V.— The valve opening outwards at the upper end of 
 the pump-head, 
 w.—Winch. 
 
 y.— Three-way communication to atmosphere, or 
 auxiliary pump. 
 
 Z.— Return-tube for mercury. 
 
 Sixty years elapsed before a second form of 
 mercurial pump was devised, by Dr. Joseph 
 Baader,* who, about 1784, devised the pump 
 depicted in Fig. 3. This pump consisted of a 
 
 Fig. 3. 
 
 hollow ellipsoidal vessel, A, at the summit of 
 the barometric column, B, terminating above in 
 a passage containing a three-way tap below it, 
 bent round into a chamber with an outlet-tap, 
 and having a second rigid tube, inclined at ar> 
 angle, ending above in a funnel-shaped vessel 
 
 * Baader. See Hiibner’s “ Physikalisches Taschenbuch.” 
 Salzburg, 1784, p. 630, referred to in Gehler’s “Physikalisches 
 Wbrterbuch,” vi,, p. 602 (Ed. 1831), 
 
4 
 
 to receive the mercury. To work this pump, 
 the bottom tap was closed, and the three-way tap 
 was turned so as to make the pump-head, A, 
 communicate with the outer air. Mercury was 
 then poured into S until A was filled ; the three- 
 way tap was then turned to cut off communica¬ 
 tion with the air, and to connect A with the re¬ 
 ceiver or other vessel to be exhausted. The 
 bottom tap was then opened, and a portion of 
 the mercury permitted to run out until the top of 
 the barometric column was below A, producing 
 a partial vacuum in the pump-head and the 
 receiver. The lower tap was then closed, the 
 three-way tap returned to its former position, 
 and the mercury which had flowed out of 
 the pump was poured again into S, and 
 the operation repeated. It will be obvious 
 that this operation is equivalent to raising and 
 lowering the supply-vessel, as in the pumps of 
 Geissler and Toepler, and their prototype of 
 Swedenborg. Baader had a curious history. 
 In 1784, when this pump was devised, he was 
 studying in Vienna. He afterwards went to 
 
 Fig. 4. 
 
 Gottingen, "thence to Edinburgh, where he 
 took the degree of Doctor of Medicine ; but 
 he quitted the medical profession for that of 
 mining, and was afterwards connected with 
 an iron works at Wigan. Gehler states that 
 Baader abandoned his first form of pump in 
 favour of a second,* shown in Fig. 4, which 
 
 * Baader. See Hiibner’s “Taschenbuch,” cited above; 
 referred to, with figure, in Gren’s “Journal der Physik,” 
 
 differs from Fig. 3 in having the supply- 
 vessel capable of being raised and lowered, 
 not by a flexible tube, but by turning it about 
 on a joint provided at the meeting point of the 
 barometric tube and the supply tube. 
 
 A very finely-constructed pump, on Baader’s 
 principle exists in the collection of apparatus 
 in the physical laboratory of University 
 College, London, having apparently been 
 acquired in April, 1828. There is no maker’s 
 name or date on the apparatus, but it belongs 
 to, and is catalogued as being “the Torri¬ 
 cellian part” of a fine double-barrelled air- 
 pump (price ^100) by the famous instrument 
 maker John Cuthbertson, of Amsterdam and 
 London. It is shown in Fig. 5. The exhaust- 
 
 Fig 5. 
 
 pipe of brass leads from the three-way tap at 
 the top to a union joint below the plate of the 
 mechanical pump, so that the exhaustion hav- 
 
 ii., p. 326; see also Hindenburg. “ De Antlia Baaderiana,” 
 4to Leipzig, 1787,referred to in Young’s “Lectures on Natural 
 Philosophy.” A quarto volume published at Bayreuth, in 
 1797, under the title “Theorie der Pumpen,” is possibly by 
 the same Baader. 
 

 s 
 
 ■ —-- - - -■ n.-- ■ ■ in - .r—. 
 
 ing been carried in the ordinary way to a 
 certain point, it should be completed by using 
 the Torricellian part. The apparatus is 
 mounted on a massive tripod of mahogany; 
 the fittings of the taps are of metal (steel or 
 iron ; the height from floor to the top of the 
 exit funnel is just under two metres ; the height 
 from the bottom tap to the bottom of the 
 pump-head is just over i metre. The baro¬ 
 metric tube, B, is of glass, as is also the 
 supply tube. The pump-head is a substantial 
 glass vessel, almost exactly i metre in external 
 circumference, hence it would require about 
 400 lbs. of mercury to work the pump. Prof. 
 Carey-Foster states that it is believed to have 
 been acquired by Dr. Lardner’s direction ; he 
 has no knowledge of its having been used. I 
 have not been able to discover in any of Cuth- 
 bertson’s pamphlets or price-catalogues now 
 extant any reference to this part of the appa¬ 
 ratus. 
 
 A pump on the same principle was patented 
 in 1881 by Mr. Rankin Kennedy. It is de¬ 
 scribed later. 
 
 In 1787, C. F, Hindenburg* described the 
 pump shown in Fig. 6, closely resembling 
 
 Fig 6. 
 
 * Hindenburg’. See Gehler’s “ Physikalishes "Wdrterbuch,” 
 vi., p. 603 (1831 edition), which refers to Hindenburg, “ Antlias 
 novae hydraulico-pneumaticae mechanismus et descriptio.” 
 Leipzig. 4to. 1787. Young mentions also Hindenburg, 
 “ De Antlia Baaderiana.” Leipzig. 4to. 1787; and to 
 Hindenburg “ Do Antlia Nova,” Leipzig. 4to. 1789. 
 
 Baader’s, but having a piston and cylinder for 
 the purpose of driving the mercury up and 
 down in the barometric tube. 
 
 Cazalet,* of Bordeaux, states, in 1798, that 
 he has constructed an apparatus like Hinden¬ 
 burg’ s, of glass and iron, and that he has 
 succeeded with it in getting an exhaustion of 
 one millionth of an atmosphere; how he 
 measured this does not appear. A similar 
 kind of pump appears to have been used by 
 Michel.f 
 
 In 1804, A. N. EdelcrantzJ published a 
 “ Description of a Mercurial Air-pump of 
 unlimited exhausting power, with a wooden 
 piston.” This pump (Fig. 7), which appears 
 
 Fig. 7. 
 
 to have been independently conceived, is on 
 the same lines as Hindenburg’s, but with 
 superior mechanical arrangements. The three- 
 way tap, instead of being bored diagonally 
 toward the handle, is bored toward the outer 
 end, as in the modern Geissler pumps, and 
 the wooden piston (p) is moved in its stout 
 wooden casing by a rack and pinion motion. 
 
 Some years after this “ Nicholson’s Journal ’ ’ 
 contains a description of Traill’s§ apparatus, 
 which has already been mentioned. It drew 
 a further communication from R. Bancks, 
 
 * Cazalet. “ Gren’s Journal,” i., p. 478. 
 
 + Michel. See “ Gren’s Journal,” ii., p. 129. 
 
 t Edelcrantz. “ Nicholson’s Journal,” vii., p. 188. 
 
 { Traill. “ Nicholson’s Journal,” xxi., pp. 63 and 161, 
 
5 
 
 instrument maker in the Strand, that he had 
 constructed similar apparatus for Mr. Children 
 at a prior date. 
 
 In 1824, Joseph H. Patten* wrote an “ Account 
 of a New Air-pump,” which, like those of 
 Hindenburg and Edelcrantz, had a piston to 
 raise and lower the mercury, but which in 
 another point more nearly resembled the 
 original pump of Swedenborg. The pump- 
 head consisted of a wide-mouthed glass globe, 
 on the top of which was fixed a flat metal 
 plate with a raised rim. In the centre of this 
 was adopted a metal pipe, having a valve 
 opening downwards below it. This valve, a 
 mere metal plug, was held upon a stiff wire 
 which entered the tube above, and was con¬ 
 tinued downward to the summit of the baro¬ 
 metric tube, where it ended in a ball float. 
 When the mercury rose it lifted the ball and 
 closed the valve, and maintained it closed till 
 it once more fell. The air in the pump-head 
 was expelled through a separate valve opening 
 outward in another part of the plate. Patten 
 suggested that, instead of the stiff wire, ball, 
 and valve, the exhaust tube should be pro¬ 
 longed downwards to near the bottom of the 
 globe. The publication of this description 
 drew from Professor J. F. Danaf some remarks 
 suggesting, apparently without knowledge of 
 prior inventors, that it would be preferable to 
 use a three-way tap. To this Patten replied 
 in a later issue in a strongly personal vein. 
 
 A more complex form of pump, in which the 
 mercury was moved by air-pressure, applied 
 by pouring mercury into a second vessel con¬ 
 taining air—somewhat on the plan of Hero’s 
 fountain—was described in 1824 by Dr. E. 
 Romershausen,^ and in 1825 by Uthe.§ In 
 each case a complicated six-way tap was 
 requisite. 
 
 In the same year a pump by Oechsle|| was 
 described, having a three-way tap connecting 
 the pump-head alternately with the air and 
 with the receiver to be exhausted, and pro¬ 
 vided also with a barometer gauge beneath 
 the receiver. The mercuiy was lowered and 
 raised, as in Edelcrantz’s pump, by a large 
 
 * Patten. “Silliman’s Journal,” viii., p. 144, 1824, and 
 ix., p. 92, 1825. An almost identical apparatus, called an 
 “ oil air-pump,” by Sadler, is described by Gchler, vi., p. 606. 
 
 t Dana. “ Silliman’s Journal,” viii., p. 275. 
 
 $ Romershausen. “ Neue hydrostatiscbe Luft - pumpe.” 
 “ Kastner’s Arcbiv fur Naturlehre,” ii., p. 359, 1824. 
 
 ? Uthe. “ Dingler’s Polyteclmisches Journal,” xvii., 
 p. 272, July,1825. 
 
 |] Oechsle; see Wucherer: Beschreibung einer grossen 
 Quecksilbernen Luftpumpe, welche sich in dem physika- 
 lischem Cabinet in Karlsruhe befindet, “ Kastner’s Arcliiv,” 
 v., p. 329, 1825. 
 
 - --- -- -- - — - ' ' 
 
 wooden piston, but worked by a chain and 
 winch. 
 
 Next in strict historical order is the very 
 remarkable pump of Mile, 1828, which, how¬ 
 ever, belongs to the third class of machines. 
 
 K. T. Kemp,* of Edinburgh, in 1830. de¬ 
 signed a remarkable double-acting pump. In 
 this machine a winch, manipulated like that 
 of the ordinary mechanical air-pump, moves 
 two pistons up and down in two cylinders 
 (Fig. 8). These are full of mercury, which is 
 
 Fig. 8. 
 
 thereby driven alternately into the two ex¬ 
 hausting-chambers, and which serve as pump- 
 heads. They are provided with valves to 
 admit air from the receiver, and also with 
 valves opening outwardly to let out the ex¬ 
 pelled air. The latter pair of valves are in¬ 
 geniously provided with floats, so that they 
 close while there is still some mercury remain¬ 
 ing above the actual valve to seal the joint 
 and make it air-tight. The double-pump of 
 Gardiner described below (Fig. 17), closely 
 resembles that of Kemp. 
 
 The year 1855 brings us to the famous 
 pump of the late Dr. H. Geissler, of Bonn, 
 which in his hands did such excellent work 
 for exhausting the Geissler vacuum-tubes* 
 The first public mention of this pump appears 
 to be in a pamphlet entitled, “ Ueber das 
 geschichtete elektrische Licht ” published in 
 
 * Kemp. | Description of a new Torricellian Air-Pump. 
 “ Edinburgh Journal of Natural and Geographical Science,” 
 ii., p. 95,1830. See also in Fechner’s “ Repertorium des 
 Experimentalphysik,” i., p. 116 ; and Taf. 2, Fig. 13,^14, 1832. 
 Also “B.rnmg. und Ett. Zeitschr.,” viii. 193. 
 
7 
 
 Berlin in 1858* by Dr. W. H. Theo. Meyer. The 
 form at first given to this apparatus is shown 
 in Fig. 9; which shows the vessel containing 
 
 Fig. 9. 
 
 
 the supply of mercury, S, connected to the 
 lower end of the barometric column, B, by a 
 flexible india-rubber tube, g. The pump- 
 head, A, is provided at the top with a large 
 three-way tap, T. The operation of this 
 simple contrivance is obvious : the air in the 
 pump-head is repelled by raising the supply- 
 vessel, whilst the tap is in such a position that 
 the pump-head communicates with the outer 
 air. The tap is then turned so that the com¬ 
 munication with the outer air is cut off, and 
 communication is established with the vessel 
 to be exhausted. This being done, the supply 
 vessel is lowered down, when the mercury in 
 the pump-head sinks, and draws in air through 
 the exhaust tube. The tap is then turned 
 again, and the supply vessel is again raised 
 to expel the air that has been drawn in, then 
 the tap is turned and the supply vessel again 
 lowered. By repeating these operations a 
 sufficient number of times, the air remaining 
 in the apparatus is reduced to a very small 
 portion of its original amount. The per¬ 
 fection of the vacuum is limited by two causes : 
 the inherent imperfection in the three-way 
 tap, and the impossibility of expelling the 
 film of air which adheres to the inner surface 
 
 * Meyer. “ Ueber das gescbichtete elektriscbe Licht.” 
 1858. The figure in this pamphlet strikingly resembles 
 Baader’s pump. Fig. 4., p. 4. 
 
 of the pump-head. However perfectly the 
 tap may be ground into its seat, however 
 carefully it is lubricated with stiff grease, 
 still air-films will remain between the working 
 surfaces. There is a tendency for a channel 
 to be formed in the film of grease at that part 
 of the conical periphery of the tap where the 
 apertures into it slide round against the in¬ 
 ternal wall of the glass seat; and through 
 this channel in the glass minute bubbles of 
 air force their way. Moreover, the grease 
 itself may give off vapours which spoil the 
 perfection of the vacuum. The Geissler pump 
 is, moreover, subject to other defects, amongst 
 which may be enumerated the heavy labour of 
 raising and lowering the supply vessel, the 
 liability to accident to the glass top, the 
 liability to fracture of the pump-head. The 
 latter accident occurs not unfrequently, if, 
 after a certain degree of exhaustion has been 
 attained, the supply vessel is suddenly raised, 
 for then, there being little or no air in the 
 pump-head to serve as a cushion, the mercury 
 rising in the pump-head strikes with a sudden 
 blow against the upper portion of the pump* 
 head and fractures it. The long barometric 
 tube or shaft of the pump is also liable to 
 fracture. The working of the tap, being done 
 by hand, requires continuous attention; lastly, 
 the process of exhaustion is slow. 
 
 The subsequent improvements relate to the 
 removal of one or other of these drawbacks. 
 Some of them were introduced by Dr. Geissler 
 himself, and by his successor, Herr F. Muller, 
 of Bonn. The form depicted in Fig. 9 (taken 
 from Toepler’s paper in “ Dingler’s Poly- 
 technisches Journal,” clxiii., p. 426, 1862) 
 
 shows the kind of three-way tap (a modifica¬ 
 tion of Senguerd’s) originally used, in which 
 the passage for the expulsion of the air was 
 carried through the conical body of the tap to 
 its end. In another form, the same kind of 
 tap was used, but the exhaustion was made 
 through this longitudinal passage, the ex¬ 
 ternal tubular seat being prolonged and con¬ 
 nected with the vessel to be exhausted. In 
 this case the air was expelled upwards 
 through the transverse passage through the 
 tap. In yet a third form the exhaust tube was 
 sealed on at right angles to the external 
 barrel, and the tap was pierced (on Babinet’s 
 plan) with a three-way transverse passage. 
 In the most recent form given to the Geissler 
 pump by its makers, there are three taps, one 
 of them, a large one, being a three-way tap 
 of the last-mentioned kind, simplified by 
 having only one transverse bore of conical 
 
 B 
 
8 
 
 form. The arrangements of this pump are 
 shown in Fig. io, in which Ti is the three-way 
 tap, and T2 and T3 plain taps of a smaller 
 size. The use of the two upper taps is to 
 enable the last traces of air to be more per- 
 
 Fig. 10, 
 
 fectly expelled from the pump-head. During 
 the ordinary working of the pump the two 
 top taps are always left open. After the ex- 
 
 Fig. 11. 
 
 haustion has reached a sufficient point, the 
 large tap is turned so as to cut off communi¬ 
 cation with the exhaust tube, and to open 
 communication from the pump-head upwards 
 through all the taps. The supply vessel is 
 
 then raised so high as to drive the mercury up 
 above the level of the topmost tap. The top 
 tap is then closed, and, by sinking the supply- 
 vessel, the mercury is caused to fall below the 
 pump-head. Itjs then slowly raised, driving 
 before it all the air that may remain in the 
 pump-head, and collecting it just below the 
 top tap. As soon as the mercury has risen 
 through the second tap, this is then closed, 
 and the mercury once more lowered. It will 
 be seen that if the space between t 2 and T 3 is 
 sufficiently great, this residual air will not be¬ 
 come compressed to anything like atmospheric 
 pressure, and hence the air-films forming in 
 the upper part of the pump-head or in the 
 channels of the three-way tap will be com¬ 
 paratively slight. The curved tube at the 
 top is used for collecting the gases extracted 
 in chemical operations ; the horizontal ex¬ 
 haust tube is usually connected with drying- 
 vessels, and with appropriate gauges. The 
 complete pump as used for the purposes of the 
 chemical laboratory as shown in Fig. 11. 
 
 Morren’s pump* * * § is practically identical with 
 Geissler’s, save in a detail concerning the 
 three-way tap, which has, instead of having a 
 glass cone with a longitudinal passage pierced 
 through it, a steel cone, with a channel or 
 groove cut in the side. Another almost iden¬ 
 tical form was described by G. Jolly,f about 
 the same time. It resembled Geissler’s first 
 form, but was provided with a mechanica 
 device of a winch and pulley, to raise and 
 lower the supply-vessel. This device was 
 speedily adopted by other makers ; it may be 
 seen in the latest form of Geissler’s pumps. 
 Jolly’s pump also has a steel three-way tap. 
 Von BaboJ sought to replace the egg-shaped 
 pump-head by a glass cylinder, strongly 
 held together between circular steel ends, 
 and having two automatic valves opening 
 into and out of it, precisely as in Kemp’s 
 pump of 1830 (Fig. 8). Poggendorff§ further 
 improved the mechanical lift by adding 
 a counterpoise weight. Alvergniat|| intro¬ 
 duced an automatic valve into the exhaust 
 tube, to prevent the mercury from being driven 
 
 * Morren. “ Annales de Chim. et de Phys.” March, 1865. 
 See also, “Carl’s Repertorium der Physik,” vol. i. p. 142, 
 1866. It is depicted in Ganot’s “ Physics” (1879), p. 158. 
 
 + Jolly. Ueber eine neue Einrichtung der Quecksilber- 
 Luft-Pumpe. “ Carl Repert,” i. p. 144. 1866. See also 
 Muller-Pouillet’s “Physik” (ed. 1876), vol. i., p.231; and 
 Mousson’s “Physik” (ed. 1879), i., 1878. 
 
 t Yon Babo. See “ Miiller-Pouillet,” i. 233. 
 
 § Poggendorff. “ Pogg. Ann.,” exxv., 151. 
 
 II Alvergniat. See Pellat, “ Cours de Physique,” (1883) 
 
 3 i 9 - 
 
9 
 
 back into the exhausted vessel. Weinhold * 
 modified the glass three-way tap, and added 
 above it a small chamber, which was closed 
 by a second glass tap of simple construc¬ 
 tion. (See Fig. 37, in Appendix.) The 
 use of this upper chamber to secure a more 
 perfect- vacuum brings this form almost into 
 identity with Geissler’s later form. The device 
 of interposing such a chamber between the 
 three-way tap and the external air was inde¬ 
 pendently suggested in 1875 by Kundt and 
 Warburg,f and appears to have been already 
 adopted by GeisslerJ in 1873. 
 
 In the years 1873 and 1874, Dr. Joule § 
 devised several forms of mercurial pump, 
 having but one valve, an india-rubber plug, v, 
 fitting into a cone seat at the top of the pump- 
 head, allowing the air to be expelled. A little 
 
 Fig. 12. 
 
 mercury above it kept the joint secure. The 
 necessity of emptying a second valve was 
 obviated by the device (previously used by 
 Mile) of connecting the vessel to be exhausted 
 to the pump by means of an overhead tube, H, 
 of more than barometric height which passed 
 
 * Weinhold’s. “ Physikalische Demonstrationen ”(1883) p 
 175. See also “ Carl’s Repert,” ix., p. 78, 1874. 
 t Kundt und Warburg. “ Pogg. Ann.,” 526, 1875. 
 t See Bessel-Hagen. “ Wied. Ann.,” xii., 436, 1881. 
 
 5 Joule. “ Proc. Lit. Phil. Soc. Manchester,” xii., 57, 1873 >’ 
 tb. xiii., 58, 1874; ib. xiv., 12, 1875; see also “Scientific 
 Papers,” i., 623; also “Catalogue of Loan Collection of 
 Scientific Apparatus,’’ 1876, p. 133. 
 
 into the pump-head, its open lower end descend¬ 
 ing nearly to the bottom of the chamber. 
 This acted automatically as a trap; for, on 
 the raising of the supply-vessel, the mercury 
 rising in the pump-head first closed the orifice 
 of the head-tube, so cutting off communication 
 with the air that remained in the pump-head, 
 but never rising in that tube to a height 
 greater than 76 centimetres above the level of 
 the valve. Another feature of Joule’s pump 
 was that the supply-vessel was closed at the 
 top, being made of a globular flask. By 
 allowing only a certain amount of air to enter 
 this flask, the pressure inside was kept at less 
 than atmospheric, enabling the length of the 
 shaft, B, to be reduced. Later, a conical 
 ground-glass tube was used in place of the 
 india-rubber plug for a valve. 
 
 In 1873, Mitscherlich * altered the pump in 
 the manner shown in Fig. 13. The 
 
 Fig. 13. 
 
 double function performed by the three 
 way tap was in this form shared between 
 an automatic valve, V, opening upwards only, 
 and a plain glass tap, T, worked by hand. 
 The valve consisted of a rod of glass, ground 
 conical at its lower end, fitting into a tube of 
 one centimetre internal diameter. This rod 
 was raised from its seat by the mercury as it 
 rose through the pump-head. A perforated 
 cork placed in the eject tube above the valve 
 prevented it from rising too high, otherwise 
 it would, in falling, become jammed. The 
 
 * Mitscberlich. “ Pogg, Ann.,” cl., 420, 1873. 
 
10 
 
 exhaust tube communicated with the pump 
 through the tap, T, at a point below the pump- 
 head ; the communicating tube being enlarged 
 to receive anhydrous phosphoric acid or other 
 drying materials, the same being protected 
 from the rising of the mercury by the inter¬ 
 position of a loosely-fitting glass valve of 
 ovate shape, u. 
 
 Another modification, due to Lane-Fox,* is 
 shown in Fig. 14. The valve at the top of the 
 
 Fig. 14. 
 
 Lj 
 
 Lane-Fox’s Pump. 
 
 pump-head is a conical glass stopper ground 
 to fit tightly into its seat, requiring to be re¬ 
 moved and replaced by hand. The overhead 
 tube, H, which acts as a barometric trap, is 
 joined to the shaft of the pump just below the 
 pump-head. Lane-Fox also suggested the 
 use of an automatic valve (like that of 
 Alvergniat) to obviate the necessity of using 
 the tall head-tube. This pump was for a long 
 time used by the Anglo-American Brush Elec¬ 
 tric Light Corporation for exhausting the Lane- 
 Fox incandescence lamps ; they introduced a 
 number of modifications in detail, one of which 
 consisted in replacing the stopper, V, by an 
 automatic valve resembling Mitscherlich’s. A 
 side-tube leads from below the lower automatic 
 valve, u, to the tap of the pump-head. There 
 is also a spark-gauge. A drawing of one of 
 the intermediate forms of Lane-Fox pump is 
 given in Gordon’s “ Electric Lighting,” (1884) 
 P- 83. 
 
 About the same date, minor improvements 
 were suggested by several persons. Mr. Dew- 
 
 Smith, of Cambridge, suggested the use, at 
 the top of the pump-head, of an automatic 
 valve consisting of a strip of rubber or silk 
 stretched over an orifice, precisely as in many 
 mechanical air-pumps, the valve itself being 
 surrounded by an upper mercury cup to ensure 
 a tight joint. Messrs. Goebel and Kulen- 
 kamp,* who used an automatic glass valve to 
 close the top of the pump-head, adopted above 
 it a flexible tube, by means of which to return 
 to the supply-vessel the small quantities of 
 mercury which from time to time were driven 
 up through the pump-head, with the ejected 
 air. Guglielmo,t applying a very similar de¬ 
 vice, achieved the not unimportant result of 
 causing the tap at the top of the pump-head 
 to discharge the ejected residual air into a 
 space already partially exhausted. This he 
 accomplished by interposing in the flexible 
 tube connecting the summit of the pump-head 
 with the closed top of the supply-vessel a 
 vertical glass tube, about 20 centimetres long, 
 with a three-way tap opening also into the air. 
 Through this tap atmospheric pressure could 
 be momentarily established when the supply 
 vessel was in its lower position, and nearly full 
 of mercury. When it was raised, the mercury 
 ran out of it into the pump-head, leaving the 
 space in it partially exhausted; and into this 
 vacuous space the three-way tap at the top of 
 pump-head opened to let out the ejected air. 
 As will be seen later on, this device makes 
 this pump resemble somewhat some pumps of 
 the third group. 
 
 Mr. Albert GeisslerJ has replaced the three- 
 way tap by two automatic valves (Fig. 15, p. 31} 
 one of which, v, opens from the top of the pump- 
 head into the outer air; the other, u, admits 
 air from the vessel to be exhausted into the 
 pump just below the pump-head, as in the 
 Mitscherlich and Lane-Fox pumps. These 
 valves are hollow tubes of glass, with spindle 
 ends to guide their motion, which float in the 
 mercury when it reaches them. They are 
 provided with accurately ground glass collars 
 instead of conical ends, to fit against the ends 
 of the tubes which they respectively close. An 
 additional tap, T, is interposed for safety 
 between the pump and the vessel to be ex¬ 
 hausted. Other tubes lead to the manometric 
 gauge and to the drying apparatus. This 
 form of pump is intended for industrial use, 
 
 * “ Wied. Beiblatter,” vi., 849. 1882. See also Specification 
 of Patent 5,548 of 1881. 
 
 + Guglielmo. “Wied. Beibl,” viii., 739, 1884. 
 
 t A. Geissler. “Centralzeitung fur Optik and Mecbanik,” 
 vii., 12. 1886; also D. R. Patent, No. 32,224, 1885. 
 
 * Lane-Fox. Patent Specification 3494, of 1880. 
 
where power is available to keep the supply- 
 vessel slowly rising and sinking. A small 
 improvement of recent date, due to Messrs. 
 Greiner and Friedrichs* consists of a new 
 three-way tap of peculiar construction, pierced 
 with two transverse channels at 45°. Of the 
 three openings, two are at one side of the 
 barrel,' one at the other, so that the tap has to 
 
 Fig. 15. 
 
 be worked through 180 0 instead of 90°, and 
 the channels in the grease do not lead directly 
 from one aperture to another. Hence there is 
 a lesser risk of leakage. 
 
 In 1881, Mr. Rankin Kennedyf prepared a 
 pump for exhausting lamps, going back in 
 principle to that of Baader. Mercury is passed 
 down a supply tube, s (Fig. 16) and rises with 
 the pump-head, expelling the air through a 
 valve, v. Then a three-way eject tap, E, at 
 the bottom is turned, cutting off communica¬ 
 tion with S, and allowing all the mercury in 
 the pump-head to run out into a basin below. 
 The lamp, or other vessel, R, to be exhausted, 
 is joined in through an aperture at the top of 
 the pump-head by means of a tube passing in 
 
 * Greiner and Friedrichs. “Wied. Ann.,” xxix., 672, 
 1886; and “ La Lumiere Electrique,” xxiii. 333, 1887. 
 
 t Kennedy. Specification of Patent, 5,524 of 1881. See 
 also Dredge’s “ Electric Illumination,” ii. p. ccxxxii, 
 
 through an india-rubber cork, c, and sealed 
 above by a mercury joint. This tube is also 
 supplied with an automatic valve, v y at its 
 lower end, to allow it and the exhausted lamp to 
 be removed from the pump, in order to seal it off. 
 A similar device had also been described by 
 Akester. Akester’s pump* closely resembled 
 that of Lane-Fox ; but in it the raising and 
 lowering of the supply-vessel was obviated by 
 using, at the bottom of the pump-shaft, a 
 
 Fig. 16 
 
 Kennedy’s Pump. 
 
 three-way tap, enabling the barometric column 
 to be placed alternately in communication 
 with a supply cistern placed at a high level, 
 and with a return-pipe through which the 
 descending mercury flowed away at a lower 
 level, to be again pumped up to the high level 
 by a mechanical pump. Another way of raising 
 andloweringthemercuryinthepump, which also 
 dispensed with flexible tubing, was suggested 
 by Rock f (A similar device was suggested by 
 Mile in 1830.) The pump-head and barometric 
 column are formed by a single straight cylin¬ 
 drical tube, about 100 centimetres long, 10 
 centimetres in internal diameter of glass, and 
 1 centimetre in thickness. It is open at the 
 bottom, but closed in and furnished with the 
 usual three-way tap at the top. Outside it is a 
 
 * Akester. See Specifications of Patents, 4,458 of 1881, and 
 2,519 of 1882. 
 
 t Rock. “ Wied. Beibl,” vii., 79°> 1883. 
 
second, slightly longer, tube, having an in¬ 
 ternal diameter of 12*6 centimetres. This— 
 the supply-vessel—is closed at the bottom, 
 and can be raised or lowered mechanically. 
 If the outer one, filled with quicksilver, is 
 raised, the liquid forces its way up the inner 
 tube, driving the air before it, through the 
 three-way trap. When it is lowered, the 
 mercury remains inside, to a height which will 
 never exceed 76 centimetres above the level of 
 the mercury in the outer tube, the space above 
 being left vacuous. The inventor claims that 
 this construction is less liable to give trouble 
 than the usual form. Cruto* has used a 
 somewhat similar device, but with sul¬ 
 phuric acid instead of mercury in the pump. 
 To obviate having to work with a pump- 
 shaft twenty feet long, he adopted the device 
 of an auxiliary exhaust pump. Narrf has 
 described a simple pump on Jolly’s plan, but 
 
 Fig. 17. 
 
 having steel taps, the pump-head of glass 
 being united above and below to the working 
 parts by careluKy-ground and lightly-greased 
 steel unions, clamped together by screws. By 
 
 * Cruto. Specification of Patent, 1895 of 1882. 
 
 + Narr. Ueber eine Abanderung der Jolly’schen Queck- 
 silberluftpumpe. “ Wied. Annalen,” 542, 1885. 
 
 reason of its strength, this construction seems 
 to be preferable in cases where very high vacua 
 are not required. 
 
 Double-action pumps have been suggested 
 by various persons. Kemp’s pump (Fig. 8) 
 was of this class, so is one by Serravalle,* in 
 which there are two supply-vessels, so arranged 
 that while one rises the other descends; two 
 separate pump-heads, with two three-way taps 
 automatically opened and shut; and two ex¬ 
 haust tubes uniting into one. Another double 
 pump, by Gardiner,f depicted in Fig. 17, is 
 worked mechanically from a rotating shaft. 
 Two eccentrics drive alternately up and down 
 two hemispherical pistons, Pi and P 2 , which 
 press in the flexible hemispherical bottoms of 
 the two supply-vessels. The valves of this 
 pump are all automatic, as in Kemp’s pump. 
 It is provided with a barometric gauge, G, and 
 a comparison barometer, b . 
 
 Sub-Class I#.—Shortened Upward- 
 driving Pumps. 
 
 The length of the pump-shaft in the pre¬ 
 ceding cases being necessarily equal to that 
 of the barometric column, renders all these 
 forms of apparatus more or less unwieldy. 
 Although a column of mercury 76 centimetres 
 high is a necessity for working between 
 vacuum within and atmospheric pressure with¬ 
 out, no such length is required when working 
 between vacuum and a reduced pressure. In 
 fact, the length of the pump may be shortened 
 by reducing the pressure of the air above the 
 surface of the mercury in the supply vessel, 
 in all pumps of Class I. and Class II. The 
 first suggestion for shortening the pump came 
 from the Rev. Professor Robinson,! in 1864, 
 and was almost immediately followed by one 
 from Professor Poggendorff. § In these appa¬ 
 ratus a common air-pump worked by hand was 
 used to produce a partial vacuum. The pump- 
 shaft was quite short, and ended in an 
 auxiliary chamber, closed at the top, but 
 having a tap communicating either with the 
 auxiliary pump or with the outer air. The 
 three-way tap above the pump-head also 
 opened into a tube which could be made to 
 communicate either with the auxiliary pump or 
 with the outer air. To fill the pump-head with 
 
 * Serravalle. “ Riv. Scient. Industr.,” xiv., 401, 1882 ; also 
 “ Wied Beibl,” vii., 490, 1883. 
 
 t Gardiner. See “ La Lumiere Electrique,” xiii., 219, 1884 
 
 $ T. R. Robinson. “ Description of a New Mercurial 
 Gasometer and Air-pump.” “ Proc. Roy. Soc.” xiii., 321, 
 1864. Phil. Mag., xxviii., 235, Sept. 1864. 
 
 ? Poggendorff. “ Pogg. Ann.,” cxxv. 151, 1865. See also 
 Muller-Pouillet’s “ Physik” (1876), i. 233. 
 
13 
 
 mercury, air was admitted to the auxiliary 
 chamber, whilst at the same time the auxiliary 
 pump was applied at the top to suck the mer¬ 
 cury into the pump-head. The three-way tap 
 being 1 then turned to put the pump-head into 
 the vessel to be exhausted, the auxiliary pump 
 was used to reduce the pressure in the auxiliary 
 chamber, causing the mercury in the pump- 
 head to fall, the height of the column repre¬ 
 senting always the difference between the 
 pressure in the pump-head and that in the 
 chamber. 
 
 Several later experimenters have adopted 
 this device of applying an auxiliary pump to 
 shorten the vacuum pump; and as we shall see, 
 the device is applicable to each of the three 
 main classes of pumps. Dr. F. Neesen,* 
 whose more recent pump is described later, 
 adopted this device in 1878. At that date he 
 was employing a shortened Geissler pump, to 
 which, independently of Mitscherlich, he had 
 applied an automatic exit-valve above the 
 pump-head. He had also introduced another 
 notable improvement, namely, a side-tube, 
 connecting the exhaust-tube from a point a 
 little beyond where it branched off from the 
 pump-shaft to a point above the pump^hetnT~ 
 below the automatic valve. Such a side-tube, 
 marked N in the Figs. 18 and 33, prevents the 
 fracture of the top of the pump-head by air 
 bubbles suddenly rising through the 'mercury 
 in the barometric column. \ 
 
 Schuller, f in 1881, described ^pother 
 shortened pump, with numerous carefully 
 
 # v 
 
 considered details, and a curious automatic 
 method of operating, which, however, need 
 not here be described. Fig. 18 shows 
 Schuller’s pump. V and u are automatic 
 valves consisting of small pieces of flat 
 glass, preferably triangular, which close the 
 mouths of tubes that have been also care¬ 
 fully ground flat. In the valve V, shown in 
 detail in Fig. 19, the weight of the small glass 
 plate is partly sustained by the ring of mercury 
 surrounding the tube end above which it lies. 
 
 Another feature of Schuller’s pump is the 
 valve J, situated in the tube between the pump- 
 head and the upper valve v. This valve j, 
 shown larger in Fig. 20, is composed wholly of 
 mercury, which, during the descent of the body 
 of liquid down the tube, forms, by virtue of its 
 great surface-tension, a cap over the orifice 
 three millimetres in diameter, which is here 
 interposed. As in Geissler’s later pumps, there 
 
 * Neesen. “Wicd. Ann.” iii., 608,1878 ; also “Zeitschrift 
 fur Instrumentenkunde ” ii., 287, 1882. 
 
 + Alois Schuller, “ Wied Ann.” iii„ 528, i88r. 
 
 is an auxiliary chamber, M, between v and J, 
 in which a partial vacuum is formed, so that 
 the residual air expelled from the pump-head 
 
 Fig. 18. 
 
 Schuller’s Mercury Valve. 
 
 is driven into an already exhausted space. The 
 little vault of mercury over the aperture in j 
 is able to withstand the difference of pressure 
 
H 
 
 between the partial vacuum above and the 
 nearly perfect vacuum below it. At the com¬ 
 mencement, a partial vacuum is made in the 
 pump-head, through the upper valve, by an 
 auxiliary mechanical pump. A three-way tap, 
 Y, suffices to put the space in the bottle, L, 
 alternately into communication with the at¬ 
 mospheric air, and with a tube leading to an 
 auxiliary mechanical air-pump. 
 
 Another pump of this class, by Dittmar*, has 
 simply two plain glass taps, one above, the other 
 in the exhaust-tube at the side of the pump-head. 
 It obviously could not give a good vacuum. 
 
 The most recent pump in this category 
 appears to be that of M. A. Joannis.f The 
 ordinary three-way tap at the top of the pump- 
 head communicates with the open air; below, 
 at the lower end of the pump-shaft, is a closed 
 vessel, communicating by another three-way 
 tap with a water aspirator, and with a source 
 of pressure by means of which the mercury is 
 alternately raised and lowered. 
 
 Class II.— Downward-driving Pumps. 
 
 The idea of expelling the residual air down 
 a barometric column originated with Dr. 
 Hermann Sprengel,:f who in the year 1865 
 brought out the pump which is associated 
 with his name. He had in the preceding 
 years been studying the uses in the laboratory 
 of the water-trombe or aspirator, a much older 
 instrument, used for some hundreds of years 
 for delivering air under pressure. The theory 
 of this ancient apparatus had already received 
 the attention of Magnus, § and of Buff, || and 
 Sprengel had himself devoted some attention 
 to this method of furnishing air for the blow¬ 
 pipe. It appears to have been an original 
 idea with him to substitute falling mercury in 
 the place of falling water, in order to extract 
 gases by means of the vacuum produced above 
 the column. 
 
 The form of the original Sprengel pump is 
 shown in Figs. 21 and 22. The supply-vessel, 
 S, was, in this case, a funnel fixed at the top 
 of the apparatus, from which the mercury was 
 delivered at a steady rate through a narrow 
 india-rubber tube, nipped by an adjustable 
 pinch-cock. After passing this point it fell 
 in drops down a glass tube, F, of narrow bore 
 
 * Dittmar. See “ Challenger Report : Physics and 
 Chemistry,” vol. I., 1884 ; plate 3. 
 
 + Joannis. Modification de la machine pneumatique a 
 mercure. “Ann. Chim. Phys.” Series VI., xi., 285, 1887. 
 
 t Sprengel. “Journ. Chemical Soc.,” Series II, iii., 9, 
 1865 ; see also “ Pogg. Ann.,” cxxix., 564, 1865. 
 
 ? Magnus. “ Pogg. Ann.,” Ixxx , p. 32, 1850. 
 
 II Puff. “Ann. der Cliem. und Pharm.,” lxxix., 249, 1851. 
 
 II Sprengel. “ Pogg. Ann.,” cxii., 634, 1861. 
 
 but having strong external walls, and known 
 as the fall-tube. As it fell down this tube in 
 drops, it swept out the air of the tube and the 
 air which entered from the side, each drcp 
 acting as a piston to propel the air below it. 
 To secure this action, it is essential that the 
 fall-tube should not be too wide. For rapid, 
 partial exhaustions, an internal bore of 2 to 3 
 millimetres appears to be about the best size : 
 for slower exhaustions, carried to the highest 
 degree of rarefaction, a bore of i’4 to i'8 
 millimetres appears to be preferable.* During 
 the first stages of the process of exhaustion, 
 whilst yet there is a considerable amount of air 
 in the fall-tube, the successive drops of mercury 
 
 Fig. 21. 
 
 Sprengel’s Pump (Simple Form). . 
 
 move separately down the tube, almost silently, 
 being separated from one another by the inter- 
 veningcushions of air, which as they descend the 
 tube become more and more compressed. As 
 a higher degree of rarefaction is attained 
 there is no longer a sufficient cushion of air, 
 the drops fall smartly through the vacuous 
 space with a loud metallic clinking sound as 
 they strike upon the top of the barometric 
 column, which occupies the lower thirty inches 
 or so of the fall-tube. At the bottom of the 
 fall-tube the air and mercury enter a suitable 
 vessel, K, from which, if desired, the air that 
 has been carried down the fall-tube may be 
 collected. The mercury which flows into K 
 must be periodically collected and poured 
 again into the supply-funnel at the top. In 
 
 * See Gimingham. “Journ. Soc. Chem. Industry,” III. 
 
 | 84, 1884. 
 
i5 
 
 the second form (Fig. 22), a downward bend is 
 inserted between the supply-funnel and the 
 point where the mercury begins to break away 
 into drops. This bend leads to a small cham¬ 
 ber, virtually the pump-head, from the end of 
 which the mercury falls in drops. The flow is 
 regulated by a pinch-cock, r, which can be 
 
 Fig. 22. 
 
 K 
 
 Sprengel’s Pump (Second Form). 
 
 screwed up to nip, more or less tightly, a piece 
 of flexible india-rubber tubing inserted in the 
 supply tube. In this figure there is also shown 
 a small mechanical air-pump for rapidly pro¬ 
 ducing the first partial exhaustion, and baro¬ 
 metric gauge, G, to show the degree of rare¬ 
 faction attained. 
 
 The introduction of the Sprengel form of 
 pump at once attracted a revived attention to 
 the advantages of mercurial pumps for ex¬ 
 hausting, and it was soon in the hands of many 
 experimenters. Graham* applied it to extract 
 minute quantities of gases in his researches on 
 gaseous diffusion ; Bunsenf adapted it for the 
 purpose of hastening filtration, employing a 
 form somewhat modified to admit of the use of 
 drops of water instead of drops of mercury. 
 Improvements began to be introduced in the 
 details of construction, as experience revealed 
 the imperfections. It was found that air was 
 
 liable to be carried down into the pump 
 
 - — -- - ■ ■■ - ■ - — ■ ■ 
 
 ‘"Graham. “ Journ. Chem. Soc.,’’xx., 247, 1866, See also 
 “ Pogg. Ann.,” cxxix., 563, 1866. 
 
 + Bunsen, “Ann. Chem. Pharm.,” cxiviii., 277, 1868. See 
 also “Ann. Chem. Pharm.,” clxv., 159; and “Phil. Mag., 
 xlv., 153, Feb., 1873. Compare also with water pump of M 
 W. Johnson, (( Chem. Soc. Quart. Journal,” 1852, p. 186. 
 
 through the supply-vessel at such times as 
 more mercury was poured in above ; it was 
 necessary to trap off such air bubbles to 
 prevent them from vitiating the vacuum 
 already attained. It was found necessary to 
 introduce vessels containing drying materials, 
 such as concentrated sulphuric acid, or glacial 
 phosphoric acid. The fall-tubes were found 
 to have an awkward habit of cracking and 
 breaking off just at a point 30 inches above 
 the lower end, in consequence of the hammer¬ 
 ing action of the falling drops during the 
 later stages of exhaustion. Better and more 
 reliable gauges were required to verify the 
 degree of rarefaction. Lastly, some remedy 
 was wanted for the difficulty experienced in 
 forcing down through the 30 inches of baro¬ 
 metric column the last small traces of residual 
 air in the fall-tube. The falling drops ham¬ 
 mered these residual bubbles into the mercurial 
 column below them : the air bubbles, on their 
 part, always tended to rise again into the 
 upper part of the tube ; often they would stick 
 to the wall of the glass tube, refusing to move, 
 though the mercury continued to flow past 
 them. The remedy of turning on a more 
 sudden flow of mercury to sweep them out 
 was not always successful. Indeed, when it 
 is remembered that these last residua must be 
 re-compressed to atmospheric pressure, in 
 order to expel them at the bottom of the 
 tube, it would seem strange if such a method 
 of expelling the residual air should prove 
 effectual. An improved joint, made by grinding 
 the conical end of a glass tube into a conical 
 socket at the end of another tube, and placing 
 mercury in the cup surrounding the junction, 
 was described by Dr. Sprengel. 
 
 Some improvements in the desired direction 
 were made in 1869 by Professor McLeod,* who 
 introduced two important modifications which, 
 by the year i874,f seem to have come into 
 general use, and were embodied in the pump 
 as constructed by the instrument maker, 
 Cetti, of London. The mercury, instead 
 of running direct into the pump from the 
 supply-vessel, was carried down a vertical 
 tube, surrounded by a wider external one, so 
 that any air bubbles accidentally carried down 
 escaped up the wider tube instead of entering 
 the pump; also after passing this trap, the 
 mercury was forced up again in a moderately 
 wide tube, ascending to a slightly lower level 
 than the supply-vessel, and again descending. 
 
 * McLeod. “ Journ. Chem. Soc.,” vii., 311, 1869. 
 
 t See E. J. Osmond in “English Mechanic,” xix., 372, 
 1874, giving drawing of Sprengel pump. 
 
] 6 
 
 The upper bend of this tube communicated 
 at its highest point with a small stoppered 
 chamber, partially exhausted of air by filling 
 it with mercury, which was then allowed to 
 run out. As the mercury passed over this 
 bend, it consequently fell through a partially 
 exhausted space, and was still more thoroughly 
 freed from air before ascending to the head of 
 the actual pump. These improvements are 
 embodied along with others in the form con¬ 
 structed by Alvergniat,* of Paris, who worked 
 for and with the advice of M. H. Sainte-Claire 
 Deville. 
 
 In most of the modern Sprengel pumps the 
 mercury is introduced into the pump-head by 
 a jet-tube with a narrow orifice, whence it 
 spurts in a fine stream, and falls into the 
 widened tube of the fall-tube ; in some other 
 forms it merely breaks away in drops over a 
 bend in a wider tube. This form is simpler to 
 construct. 
 
 A very important addition was the improved 
 gauge introduced by McLeod,t in which was 
 applied the principle of compressing a known 
 volume of the rarefied air or gas into a smaller 
 known volume (the ratio of the two volumes 
 being accurately known) and then measuring 
 its pressure, and so calculating backwards. 
 This method, suggested long before by Arago, 
 and employed by RegnaultJ for the purpose of 
 testing the perfection of the vacuum of a 
 barometer, may be regarded as a refinement 
 upon the method of the ‘ ‘ pear-gauge ’ ’ invented 
 by Smeaton. In Smeaton’s invention the resi¬ 
 dual air left in the pear-shaped glass vessel, 
 placed with its lower end in mercury, under 
 the receiver of an air pump was, on the resto¬ 
 ration of the external pressure, driven out of 
 the body of the pear into its narrow upper end, 
 where its volume could readily be measured. 
 In McLeod’s original gauge a globe of about 
 48 cubic centimetres was employed, opening 
 at the top into a narrow volume-tube sealed at 
 the top, and suitably graduated. This appa¬ 
 ratus communicated above with the pump, 
 and stood at the top of a barometric 
 column which was provided at its foot 
 with a flexible rubber tube and a supply, 
 vessel, by raising which (by an action 
 like that of Geissler’s pump) the mercury 
 could flow up into the gauge, and force the 
 
 * Alvergniat’s form is depicted in Violle’s “ Cours de 
 Physique” (1884), I., 947. The form commonly used in 
 Germany is given in Weinhold’s “ Physikalische Dernon- 
 strationen ” (1881), 171. 
 
 + McLeod. See “Phil. Mag.” (4) xlviii. no, 1874, and 
 “ Proc. Phys, Soc. Loud.” i. 30, 1874. 
 
 t Regnault, “ Relation des Experiences ” (1847) i. 491. 
 
 residual air in the bulb into the volume-tube 
 at the top. A neighbouring pressure-tube 
 rendered the increment of pressure (due to 
 compression) evident at a glance, and all that 
 remained to be done was to multiply this in¬ 
 crement by the ratio between the volume now 
 occupied by the residual air and volume it 
 originally occupied. McLeod showed how to 
 carry the calculation to a second approxi¬ 
 mation. (See Appendix VI.) 
 
 About the same time, Crookes* introduced 
 several improvements in detail:—A method of 
 lowering the supply-vessel to re-fill it with 
 the mercury that had run through the pump ; 
 the use of taps made wholly of platinum to 
 ensure tightness ; the use of a spark-gauge to 
 test the perfection of the vacuum by observ¬ 
 ing the nature of an electric spark in it; the 
 use of an air-trap in the tube leading up to 
 the pump-head ; the method of connecting the 
 pump with the object to be exhausted, by 
 means of a thin, flexible, spiral glass tube; 
 the method of cleansing the fall-tube by 
 letting in a little strong sulphuric acid through 
 a stoppered valve in the head of the pump. 
 In carrying out these experiments, Crookes 
 was assisted by Mr. Gimingham, whose 
 further contributions to the development of the 
 pump will presently be noticed. It was with 
 this improved pattern of Sprengel that Crookes 
 was able to carry out that remarkable series 
 of researches upon the repulsion accompanying 
 radiation, which culminated, in 1875, in the 
 invention of the radiometer, and later led to 
 the discovery of the phenomena of “radiant 
 matter.’’ 
 
 Professor R. A. Meesf described another 
 modification, the fall-tube being constructed 
 with a series of bends constituting fluid valves 
 or traps, in which the minute portions of air 
 carried down the fall-tube might accumulate 
 in order to be swept out the more effectually 
 when aggregated in larger bubbles. This 
 pump also had a peculiar automatic stop-cock. 
 
 From 1875 to 1884 a series of modifications 
 were introduced by Mr. Gimingham.Firstly, 
 the process of exhaustion was accelerated by 
 the employment of multiple fall-tubes, re¬ 
 ceiving their streams from a distributing jet 
 within the pump-head. Three fall-tubes were 
 
 * Crookes. “Proc. Roy. Soc.,” xxxi., 448, 1881; also 
 “Proc. Phys. Soc. Lond.,” i., 35, 1874. 
 
 + Mees. See “ Catalogue of Loan Collection of Scientific 
 Apparatus,” 1876, p. 131. 
 
 t Gimingham. “On a new form of the Sprengel Air-¬ 
 pump.” “ Proc. Roy. Soc.,” clxxvi , 396, 1876; and Con-, 
 tributions to Development of Sprengel Air-pump.” “Journ. 
 Soc. Chem. Indust,,” 1884, 
 
Fig. 23. 
 
 1 
 
 employed, then five, later seven ; but five 
 appears to be a preferable number. In Fig. 
 23, which embodies Gimingham’s various 
 improvements with the earlier ones of Crookes, 
 there are five fall-tubes shown. Careful ex¬ 
 periments to determine the best size of bore 
 for the fall-tubes gave the following results 
 when exhausting a vessel of 136 cubic cen¬ 
 times capacity : — 
 
 Diameter of Bore ^ 
 
 in Millimetres. 
 
 Rate of flow 
 
 in cubic 
 
 centimetres 
 
 per minute. 
 
 Total quantity of Mer¬ 
 cury (in cubic centi¬ 
 metres) requisite to 
 reduce pressure to one 
 millimetre of Mercury. 
 
 Total time required 
 
 in minutes. 
 
 2-4 
 
 83-3 
 
 2,500 
 
 30 
 
 2'4 
 
 20 
 
 1,600 
 
 80 
 
 i-8 
 
 20 
 
 0 
 
 0 
 
 35 
 
 i*8 
 
 So 
 
 1,200 
 
 24 
 
 i *4 
 
 10 
 
 1,800 
 
 180 
 
 i *4 
 
 25 
 
 2,700 
 
 120 
 
 i-i 
 
 20 
 
 4,000 
 
 200 
 
 The enormous time required in the last case 
 was due to incessant choking up of the upper 
 part of the fall-tube, owing to friction in the 
 narrow bore. The conclusion derived from 
 comparison of results with varying bores is 
 that at high degrees of exhaustion the last 
 portions of air or gas are carried out by 
 entanglement with the mercury, and not by 
 the mercury acting in definite “ pistons” to 
 sweep out the gas. With respect to the length of 
 the fall-tubes, it was found that those 39 inches 
 (1 metre) long, giving a fall of 9 inches (or 22 • 5 
 centimetres), exhausted more rapidly than tubes 
 only 33 inches (85 centimetres) long. Tubes 
 longer than thirty-nine inches were found liable 
 to fracture, in consequence of the severe con¬ 
 cussions of the mercury as it fell upon the top 
 of the barometric column. Gimingham also 
 described an improvement in the McLeod 
 gauge, making its indications at once more 
 sensitive and more reliable. Several minor 
 improvements are also mentioned ; an im¬ 
 proved vacuum tap, an improved form of air- 
 trap, a radiometer gauge, and a bulb contain¬ 
 ing crumpled gold-leaf, to absorb mercury 
 vapour. 
 
 In Gimingham’s pump the supply-vessel, S, 
 
 Gimingham’s Pump. 
 
18 
 
 communicates by a long flexible tube with a 
 forked tube, y, leading to two regulating pinch 
 cocks, ri and r 2 . The left-hand tube leads up 
 through two air-taps, t and /, to the McLeod 
 gauge, G; the right-hand tube, through two 
 air-traps, / and /, to the pump-head, A, where 
 the mercury is thrown in jets into the tops of 
 the five fall-tubes. The exhaust tube has 
 four branches, one, m, leading to the McLeod 
 gauge, one, /, leading to the barometric gauge, 
 b, and one leading through the drying-tube, D, 
 and the absorbing-tube, a, to vessel, lamp, or 
 bulb, which is to be exhausted, and one, n , 
 leading to a small radiometer to serve also as 
 a gauge. A comparison-barometer is placed 
 beside b, and a measuring rod to read off 
 barometric heights is fixed at w. The supply- 
 vessel could be lowered when empty and re¬ 
 plenished with the mercury collected in K by 
 opening the pinch-cock, r y Mr. Gimingham 
 also adopted a mechanical method of counting 
 the number of times that the supply-vessel has 
 been let down to be replenished. 
 
 Mr. Gimingham has also suggested* a 
 mechanical mercurial pump with valves. 
 
 A five-fall Sprengel pump, of simpler con¬ 
 struction, and jointed with indiarubber tube- 
 joints, is used by the Thomson-Houston Com¬ 
 pany, in their factory at Lynn, Massachusetts. 
 The Anglo-American (Brush) Company have 
 also used a modified five-fall tube in their 
 works at Lambeth. Another multiple-fall pump 
 has been patented by Mr. Donkin.j- 
 
 Dr. L. von Babo}: described an ingenious 
 method of making the Sprengel pump supply 
 itself with mercury, by the device of connect¬ 
 ing it to a water-aspirator, worked by a con¬ 
 stant stream of water. This aspirator drew in 
 air and mercury at the lower part of the pump, 
 and lifted it up through a narrow tube to a 
 height above the level of the mercury in the 
 supply-vessel. Incidentally, this method has 
 the advantage, apparently not noticed by von 
 Babo, of enabling the fall-tube to be con¬ 
 siderably shortened; it has the disadvantage 
 of exposing the mercury to water-vapour during 
 a part of its circulation. 
 
 Macaluso§ proposed the addition of a 
 Mariotte’s flask, to regulate the flow of 
 mercury, thereby avoiding the need of having 
 a movable reservoir. 
 
 * Gimingham. See “ English Mechanic,” xxxvi., 442, 
 1882. 
 
 t Donkin. Centralblatt fiir Optik und Mechanik, vii., 216, 
 1866. 
 
 t Yon Babo. “ Berichte d. naturforsch. Gesellsch. zu 
 Freiburg.” ii., Heft. 3., 1879. 
 
 S Macaluso. See “ Wied, Beibl,” iv., 516,1880. 
 
 Rood,* in 1880, described several details of 
 some novelty; an iron valve in the bottom of 
 the supply-vessel, capable of fine regulation by 
 a screw, served to determine the rate of flow of 
 the mercury. Immediately below the supply- 
 vessel, the mercury entered a vacuum bulb (see 
 Fig. 24), designed to free the mercury from air 
 
 Fig. 24. 
 
 and moisture, the mercury dropping at once 
 to its lower part, which should be level with 
 the point where the curved supply-tube joins 
 the fall-tube. This bent tube, about 20 centi¬ 
 metres long, after descending gently, ascends 
 about 4*5 centimetres ; it is made of about 
 the same diameter in the fall-tube. The fall- 
 tube, as in Mees’s pump, is provided with 
 bends. Rood also described a modification of 
 the McLeod gauge. His pump was so 
 mounted that all parts of it could be heated by 
 means of a Bunsen’s burner. Great import¬ 
 ance was attached by this experimenter to 
 this point; as it appeared that much higher 
 exhaustion was thereby attained. 
 
 In 1882, Hannayf proposed to replace the 
 mercury by a fusible alloy of lead, bismuth, and 
 tin, melting at 94 0 ; by this means it was 
 thought that the necessary imperfections 
 
 * O. N. Rood. “ Sillim. Journ.” xx., 57, 1880 ; i.b. xxii , 90, 
 1881. See also New York Times, Nov. 19, 1880. 
 t Hanna}-, “ Philos. Magaz.” [5] xiii,, 229. 1882, 
 
*9 
 
 arising from the pressure of mercury-vapour 
 would be avoided. 
 
 The Sprengel pump employed by the Edison 
 Company (New York, 1885) in the manu¬ 
 facture of glow-lamps, has a simple fall-tube 
 cemented (as shown in Fig. 25) at its lower 
 
 Fig. 25 
 
 end into an iron tube, E, which carries away 
 the ejected air and mercury from a series of 
 such pumps. An iron supply-pipe overhead 
 feeds the pumps. Strong sulphuric acid 
 placed in a shallow glass vessel, D, is used for 
 drying. 
 
 Class IE*.— Shortened Downward¬ 
 driving Pumps. 
 
 Stearn,* in 1877, working in conjunction 
 with Swan at the problem of perfecting the 
 incandescent lamp, devised a shortened 
 Sprengel pump. It is obvious that the column 
 of mercury in the Sprengel fall-tube stands, 
 during the later stages of exhaustion, at about 
 76 centimetres’ height, simply because the 
 difference of pressure between the space inside 
 the tube and outside it is about equal to one 
 
 * Steam and Swan. “ On a new form of Sprengel’s Air- 
 pump.” Rep. British Association, 1877, p, 43. 
 
 atmosphere. By removing a portion of the 
 external pressure, the fall - tube may be 
 shortened to any desired extent. Accord¬ 
 ingly Stearn applied an auxiliary pump, not 
 at the top as Sprengel had done, to accelerate 
 the early stages of exhaustion, but at the 
 bottom ; the collecting chamber, K, being for 
 this purpose closed, and put into communica¬ 
 tion with the auxiliary pump. Steam’s pump 
 has undergone various modifications; in a 
 recent form* there are three fall-tubes of only 
 about 10 inches length, completely enclosed 
 in a partially exhausted chamber. In this 
 pump there are also means provided for carry¬ 
 ing up the mercury from the collecting vessel 
 back to an upper supply-vessel by closing 
 certain taps and opening others which admit 
 the atmospheric air. By this means an ex¬ 
 tremely small quantity of mercury is made to 
 do duty again and again ; and the exhaustion 
 is rapid, because with such short fall-tubes 
 there is less liability of the air-bubbles to stick 
 in the fall-tubes. The action of the pump is 
 made automatic by giving a periodic motion 
 to a three-way cock, which puts a lower re¬ 
 ceiving chamber alternately in connexion with 
 the atmosphere and with the partial vacuum of 
 the auxiliary pump. Stearn has embodied 
 sundry other modifications in patent specifica¬ 
 tions.f 
 
 A compact modification of Steam’s pump 
 has been devised by Mr. Swinburne,| who also 
 has tri'ed an inverted Sprengel pump. 
 
 The most recent shortened Sprengel pump 
 is that of Dr. W. W. J. Nicol, described before 
 the British Association at Manchester, in 
 1887. Its arrangements are depicted in Fig. 
 26. The principle of its automatic action is 
 identical with that of Von Babo, an auxiliary 
 water dropping air-aspirator (not shown in the 
 figure) being employed to draw in air at the 
 aperture, A, regulated by a tap. This air 
 draws up the fallen mercury in drops, through 
 the return tube, z, on the left, and returns it 
 into the supply chamber, S, at the top, whence 
 it passes downward through a rubber tube, 
 squeezed between the jaws of a regulating 
 pinch-cock, X, and rises through an air-trap, 
 /, into the pump-head. The distributor, is 
 simply a horizontal glass tube, sealed into the 
 
 * See Gordon’s “ Practical Treatise on Electric Lighting,” 
 1884, p. 65, giving an excellent picture. 
 
 t Stearn. Specification of patent, 5,000, °f 1881. See 
 also Dredge’s “Electric Illumination,” II. p. ccclxiv. 
 Steam’s shortened Sprengel pumps have been now for 
 several years furnished to the public by Messrs. Mawson & 
 Swan, of Newcastle-on-Tyne. 
 
 $ Swinburne, Electrician , xix., 72, 1887. 
 
20 
 
 pump-head, and pierced with small holes 
 above the openings of each fall-tube. (This 
 form of distributor originated independently 
 with Mr. J. T. Bottomley and with Mr. 
 Proctor.) The fall-tubes, f F F, are connected 
 to the pump-head in the following manner. 
 Below the pump-head short pieces of glass 
 tube of at least five millimetres bore are 
 sealed on. These are provided with small 
 flanges, and drawn out conical below, so that 
 
 Fig. 26. 
 
 they can be pushed very tightly through small 
 india-rubber plugs, which are firmly fixed 
 in mercury cups. These mercury cups, which 
 are strangulated, so as to nip the rubber 
 plugs, are sealed to the fall-tubes. The lower 
 ends of the fall-tubes pass into the collecting 
 vessel, K, through simple packings, s s s, of 
 rubber-tube. The arrows show the course of 
 the mercury. The tube, d, leads to the drying 
 
 apparatus, and to the vessel to be exhausted. 
 This pump can, of course, be used for exhaust¬ 
 ing only, not for collecting the gas for analysis. 
 The entire height of the apparatus is less than 
 one metre. A very small quantity of mercury 
 — only 300 cubic centimetres—is required 
 The entrance of water-vapour at S or a is 
 prevented by the use of tubes containing 
 calcium chloride. These pumps are now manu¬ 
 factured for sale by F. Muller, successor to 
 Dr. Geissler, of Bonn. 
 
 Class III. — Upward and Downward 
 Driving Pumps. 
 
 The earliest example of a pump which 
 drives the air up one barometric column and 
 down another, is the remarkable pump de¬ 
 vised by Professor J. Mile,* of Warsaw, in 
 1828. This pump is described by its inventor 
 as a hydrostatic air-pump without cylinders, 
 taps, lids, or stoppers. The description, as 
 will be seen by Fig. 27, is literally true. The 
 
 Fig. 27. 
 
 * Mile. Neue hydrostatische Luftpumpe ohne Kolben, 
 Haehne, Kappen, und Stopsel. <( Dingler’s Polytechnischea 
 Journal,” xxx., i., 1828 
 
21 
 
 mercury is raised in the barometric tube, B, 
 and pump-head, A, by lifting an external 
 cistern, S, of mercury by means of a winch, 
 W. The rising of the mercury first cuts off 
 communication with the vessel to be exhausted 
 by entering the mouth of the exhaust-tube, 
 which is sealed in through the pump-head; 
 and on further rising it expels the enclosed 
 air through a narrow tube, F, sealed in at the 
 top, which bends over to the right and termi¬ 
 nates below in a cup of mercury into which 
 its open end dips. This exit-tube and cup 
 constitute a barometric trap, for w r hen the 
 supply cistern is lowered the air cannot return, 
 the mercury rising in the tube, F, to a height 
 depending upon the degree of internal rare¬ 
 faction. To prevent the mercury from being 
 forced into the vessel that is to be exhausted, 
 the exhaust tube is prolonged overhead to a 
 height exceeding that of a barometric column. 
 The total height of this pump is, therefore, 
 necessarily, about nine feet. It maybe looked 
 upon as a sort of Swedenborg pump, the two 
 valves that open inwardly and outwardly into 
 the pump-head being replaced by barometric 
 air-traps. With such a pump, properly used, 
 a fairly high degree of exhaustion ought to be 
 possible. Strange to say, this pump appears 
 to have fallen into utter oblivion, and its 
 useful features have been several times re¬ 
 invented.* 
 
 The use of a second barometric column, 
 down which the air is expelled from the pump- 
 head, is generally attributed to Professor 
 Toepler,f of Dresden, whose form of pump 
 is shown in Fig. 28. Save in the use of a 
 flexible rubber-tube, and in the manner of 
 bringing the exhaust tube to the lower side of 
 the pump-head, this is identical with Mile’s 
 pump. A pump of similar form is sometimes 
 attributed to Mendeleeff; but the writer has 
 been unable to verify the reference. This 
 pump has many of the advantages and dis¬ 
 advantages of the Geissler form of pump. It 
 requires either the tall overhead tube or else 
 an automatic valve. The exit tube, F, is more 
 liable to fracture than any part of the Geissler 
 pump. But, as there are no taps to get out of 
 order, a higher degree of exhaustion can be 
 attained than with any three-way tap arrange¬ 
 ment opening into the outer air. There is 
 no need even for any other gauge than 
 
 * See, for example, H. Sutton, in English Mechanic , xxxi, 
 1882, as well as Toepler and Mendeleeff. 
 
 + Toepler. Ueber cine einfache Barometer-Luftpumpe 
 ehne Hahne, Ventile, und Schadlicben Raura. “ Dingler’s 
 Polytecbnisches Journal,” clxiii. p. 426, 1862. 
 
 the pump itself; for, as Toepler* has shown, 
 the degree of exhaustion can be measured 
 (as in the McLeod gauge) by raising 
 the mercury in the pump-head to a marked 
 point on the narrow tube just above the pump- 
 head, so as to compress the residual air into 
 the top of the narrow exit-tube, and then 
 reading off the volume and the pressure of the 
 same, and making the required calculations. 
 It possesses this obvious advantage, that the 
 last residua of air in the pump-head are swept 
 
 Fig. 28. 
 
 down the tube, F, by the mercury that falls over 
 the bend —“ Sprengelised ” over, one might 
 almost say. In fact, if it were not the case 
 that this pump antedates Sprengel’s, one 
 would be disposed to regard it as a combina¬ 
 tion of the Geissler and Sprengel pumps. 
 
 The Toepler form of pump has received in 
 recent years various modifications. E. Wiede- 
 mannf altered the overhead-tube, H, by joining 
 
 * Toepler. “ Sitzungsber. cl. Naturwiss. Gesellsck. Isis in 
 Dresden,” 1877, p. 135. See also Bessel-Hagen, “ Wied. 
 Ann.” xii., 434, 1881. 
 
 t Wiedemann. “Wied. Ann.,” x., 208, 1880, 
 
22 
 
 4 ^- 
 
 it at its base with two of Gimingham’s air¬ 
 tight joints, allowing it to be removed to be 
 cleaned.. Neesen* added the side-tube shown 
 at N in Figs. 18 and 33, to prevent the top of 
 the pump-head being broken off by violent 
 uprushes of mercury in the large bulb. Gugli- 
 elmof ingeniously connected the closed top of 
 the collecting vessel (into which the exit-tube 
 discharged air and mercury) with the closed 
 top of the supply vessel, so that as the latter 
 was raised, and the mercury ran out of it, the 
 air-pressure upon the lower end of the baro¬ 
 metric column in F was automatically lessened. 
 Improved forms of overhead-tube were sug¬ 
 gested by von HelmholtzJ and by Schuller,§ 
 and a very similar device was used by the 
 writer in 1882 to connect a glow-lamp to the 
 Lane-Fox pump by merely sealing it to the 
 top of a long barometric tube, which slipped 
 on over the top of the open overhead tube—or 
 of a tube connected with it—and dipped into 
 an external ring of mercury in a cup forming a 
 barometric air-tight trap. Neesen|| designed 
 a double-acting pump on this plan, with two 
 pump-heads, and two fall-tubes, the mercury 
 being mechanically driven alternately from 
 one pump to the other by a piston working in 
 a cylinder. The model has not yet been 
 actually constructed. 
 
 Other improvements have been made in 
 detail by Couttolene,^ who drives the air into 
 a partially exhausted space, by Diakonoff,** by 
 Bessel-Hagen,ft and by Karavodine.Jj: The 
 latter interposes between the top of the pump- 
 head and the exit-tube, F, a small chamber 
 closed at the bottom by a valve consisting 
 only of mercury standing over a capillary 
 orifice, exactly resembling that previously 
 described by Schuller (Fig. 20). This has 
 the result of causing the last portions of 
 residual air to be expelled into a space con¬ 
 taining a moderately perfect vacuum. This is 
 a decided improvement. For as was pointed 
 out with respect to the Sprengel pump, the air 
 carried down the narrow fall-tube is neces¬ 
 sarily compressed in order to drive it down 
 against atmospheric pressure, and bubbles or 
 films remain adherent to the glass. 
 
 * Neesen. See below. 
 
 t Guglielmo. “ Wied. Beibl.,” v., 16, 1881. 
 
 t Von Helmholtz. See Bessel-Hagen, “ Wied Ann.,” 
 xii., 429, 1881. 
 
 ? Schuller. “Wied. Ann.,” xiii., 533, 1881. 
 
 || Neesen. “ Wied. Ann.,” xi., 522, 1880. 
 
 IT Couttolene. “ Comptes Rendus,” xci., 920, 1880. 
 
 ** Diakonoff. See Karavodine. 
 
 t+ Bessel-Hagen, loc. cit. 
 
 tt Karavodine. “Journal de Physique,” S. II,, vol. ii.> 
 558, 1883. 
 
 In some modern modifications this is to a 
 very large extent obviated, by the device of so 
 bending the eject-tube or fall-tube, that the 
 air expelled from the pump-head need only 
 descend a very few centimetres down the tube 
 before it enters a chamber that is partially ex¬ 
 hausted. In short, if by any device—whether 
 by placing it at the top of a fall-tube or by 
 applying a good mechanical pump—a mode¬ 
 rately good exhaustion can be maintained 
 with a chamber such as that marked M, and 
 this chamber is joined to the pump-head by 
 a descending fall-tube, the length of this fall- 
 tube need not exceed the height of a column 
 of mercury representing the difference of 
 pressures between the two chambers. In the 
 diagrams that follow', such a shortened fall- 
 tube is marked Q; it may be regarded as a 
 sort of siphon air-trap. Such an arrangement 
 has been independently devised by several 
 persons. It was patented by Siemens and 
 Halske* in Germany in 1884. Fig. 29 shows 
 
 Fig. 29. 
 
 Siemens’s Pump (first form). 
 
 the device as originally designed. The pump- 
 head terminates in a capillary tube, which 
 turns over into a pool of mercury in the base 
 of the upper chamber, M, into which the residual 
 air is driven with a very slight compression. 
 When a certain amount has thus been col¬ 
 lected, it is expelled by further raising the 
 
 * Siemens and Halske. D. R. Patent, 28,579, Jan., 1884. 
 For the accompanying sketches of the pumps the writer is 
 indebted to Herr von Hefner Alteneck. 
 
mercury and opening the top tap, T, which is 
 otherwise kept closed. A wider tube, z, which 
 should be usually closed by a tap, serves to 
 return to the pump shaft the mercury which 
 
 Fig. 30, 
 
 may have been driven over into M. A later 
 form of this pump, depicted in Fig. 30, is used 
 in Siemens and Halske’s lamp factory in 
 Berlin. 
 
 A similar device was suggestedby Sundell,* 
 who has further improved the arrangements at 
 the bottom of fall-tube, so as to allow of other 
 gases being admitted to the pump. Some of 
 Neesen’s pumps, and that used in the Weston 
 lamp factory in Newark, N.J., also have this 
 device; but these belong to the sub-class of 
 shortened pumps, and are described below. 
 
 Mr. Swinburne,f who has had extensive ex¬ 
 perience with pumps of several kinds, has 
 described a form in which this principle is 
 applied. Swinburne’s first form, though pro¬ 
 vided, like Toepler’s, with a fall-tube, had 
 also an automatic valve above the pump- 
 head. Fig. 31, adapted from Swinburne’s paper 
 in The Electrician , shows this valve situated 
 above a small cavity, c, separated from the 
 pump-head by a constriction, the object of 
 which is to prevent the glass bottom of the 
 valve being broken by the sudden rise of the 
 mercury. The eject chamber, K, is connected 
 
 * Sundell. “ Wied. Beibl.,” ix , 193. 1885. 
 
 f Swinburne. “ The Electrician,” xix., pp. 51, 71, 117, and 
 158, 1887 ; a series of papers giving a summary of valuable 
 experience in exhausting glow-lamps. 
 
 through a tap, T, to a horizontal pipe. 
 This pipe, which runs along a whole range 
 of pumps in the pump-room of the lamp- 
 factory, is mechanically exhausted, and the 
 use of the tap, T, is to start the action of 
 
 * Fig. 31. 
 
 Swinburne’s Pump 
 
 the pumps. After this the action is kept 
 goingby a three-way tap (here marked y) which 
 connects the cavity above the mercury in the 
 vessel, L, alternately with the atmosphere and 
 with a supply of compressed air. In a later form, 
 (Fig. 23) the pump has a siphon mercury-trap, 
 Q, between the pump-head and the automatic 
 valve v, above it. When the exhaustion has 
 been carried far enough, the mercury is lowered 
 and raised some ten or twenty times, just so 
 
 * This figure is kindly lent by the Editor of The Elec* 
 trie tan. 
 
H 
 
 far as to drive the residual air through the 
 mercurial siphon, which then will show a 
 small back-pressure—perhaps of only one or 
 two centimetres. If the volume of the pump- 
 head is many times as great as that of the cavity 
 beyond the mercury-trap, and if there be a 
 fairly good vacuum beyond the trap, it is 
 obvious that a back-pressure of one or two 
 
 Fig. 32. 
 
 centimetres as the result of twenty strokes may 
 mean a very high degree of exhaustion. 
 Swinburne remarks that the bore of the siphon 
 tube used as a trap must be not larger in the 
 descending part than in the part that ascends 
 to the supplementary chamber. 
 
 Class IIIa.—Shortened Upward and 
 Downward Driving Pumps. 
 
 Swinburne’s pump just described might, if 
 worked intermittently with an exhausting in¬ 
 stead of a compressing pump, be transferred 
 to the category of shortened pumps. 
 
 Probably the most perfect of pumps in this 
 class is that of Prof. F. Neesen,* of Berlin. 
 This indefatigable worker has introduced, 
 from time to time, several improvements. As 
 mentioned above, he introduced the side-tube, 
 N, in 1878, and designed a double-acting 
 Toepler pump in 1880. Independently of 
 Mitscherlich, he introduced the automatic 
 valve above the pump-head. In 1882 he was 
 already employing the recurved siphon trap, 
 Q, between the pump-head and the second 
 
 * Neesen. “ Wied. Ann,” iii., 608, 1878 ; ib., xi., 522,1880 ; 
 ib., xiii., 384, 1881; “ Zeitschr. fur Instrumentenkunde,” ii., 
 287, 1882 ; ib., iii., 245, 1883 ; also, “ Wied. Beibl.,” vii., 651, 
 1883. Figs. 33 and 34 are from sketches kindly furnished] by 
 Prof. Neesen. 
 
 chamber, M. His complete pump, as con¬ 
 structed in 1887, is shown in Fig. 33. The 
 lower portion is constructed on Robinson’s 
 plan, air-tight connections being formed 
 at the three necks of the bottle, L, by 
 the use of coned steel collars, that are 
 cemented to the three tubes, and fit to 
 coned adapters, cemented to the three necks. 
 Steel screw-caps clamp down the conical 
 collars into their respective seats. The tube, 
 Y, is put into alternate communication with 
 the atmosphere, and with a good mechanical 
 air-pump, so as to raise and lower the mercury 
 
 Fig. 33. Fig, 34. 
 
 Neesen’s Pump, 1887. (Form of Valve.) 
 
 alternately in the pump-head, A. There is an 
 automatic valve, U, in the exhaust-tube, which 
 leads up to the drying flask and to the lamp or 
 other vessel that is to be exhausted. This 
 valve, which is shown enlarged in Fig. 34, is 
 made somewhat on the plan of Schuller, 
 described above, with a small glass disk about 
 two centimetres in diameter, cut from thin 
 plate-glass, which, as the mercury rises under 
 it, is pressed up against a flat flange, fashioned 
 on the lower end of the upper tube. It works 
 in a manner that leaves nothing to be desired. 
 
This pump is further provided with a chamher, 
 
 M, and a siphon-trap, Q, down which the 
 
 residual air from the pump-head is expelled 
 
 into a moderately perfect vacuum. 
 
 Another very interesting and extraordinary 
 
 pump belonging to this class is that of P. 
 
 Clerc, depicted in Fig. 35. The apparatus 
 * 
 
 Fig. 35. 
 
 1 
 
 ! 
 
 hown is connected by a flexible rubber tube 
 to a mechanical pump capable of giving a 
 moderately perfect vacuum. The apparatus 
 consists of a disk of wood, round the periphery 
 of which is fix:ed a glass tube, closed in itself, 
 but provided with a U-shaped bend to serve 
 as an air-trap. At one side of this trap rises 
 a short branch tube to which the lamp that is 
 to be exhausted is sealed; at the other a 
 similar branch tube leads to a bulb connected 
 through a tap to the auxiliary pump. Enough 
 mercury is placed in the tube to occupy about 
 a quarter of the circumference and fill the 
 trap. The whole apparatus is mounted 
 obliquely upon another disk of wood, in such 
 a way that it can be rolled round on its 
 periphery by means of a projecting central 
 handle. A preliminary exhaustion having 
 been attained, the tap is closed and the 
 apparatus is rolled around. The mercury in 
 the tube sweeps the air before it into the bulb, 
 and, passing into the trap again, emerges to 
 push a fresh quantity from the lamp in front 
 of it, leaving behind every time in the trap 
 a sufficient quantity of mercury to balance 
 the difference of pressure between the bulb 
 and the lamp. The quantity of mercury 
 required for this apparatus cannot exceed a 
 few cubic centimetres at the most. 
 
 Fig. 36! represents the pump used in the 
 lamp factory of Mr. Weston, at Newark, New 
 Jersey. The second chamber, M, connected 
 with the pump-head by the siphon-tube, Q, 
 will at once be recognised, as also the auto¬ 
 matic valve, U, in the exhaust 'tube. The 
 
 * Clerc. “Dingler’s Polytechnisches Journal,” cclxii., 
 Part II., 1886; also “ Zeitschrift fur Instrumentenkunde,” 
 vi , 403, 1886; and D. R. P. 36447 of 1885. 
 
 t For this sketch the writer is indebted to Prof. G. Forbes. 
 
 mercury in the supply-vessel, s, is raised and 
 lowered by alternately connecting the upper 
 part of the vessel through the three-way tap, 
 Y, with a mechanical exhaust pump, and with 
 the atmosphere. The top tap is only used 
 when the chamber, M, has to be put into 
 
 Fig. 36. 
 
 communication with the mechanical pump ; 
 the other taps are safety taps, not used during 
 the working of the pump. The tap between 
 the lamps and the valve, U, is worse than 
 useless. 
 
 Class IV.— Combination Pumps. 
 
 It has been suggested by Edison* and by 
 Bohmf to combine a Geissler pump with a 
 Sprengel pump in the endeavour to obtain a 
 more perfect result. This method of com¬ 
 bination, which consists merely in sealing the 
 exhaust tubes of each pump together and to 
 the lamp, cannot be commended. If the 
 Geissler exhausts more perfectly than the 
 Sprengel, or vice versa , then the other pump 
 is useless. A much more hopeful combination 
 
 * Edison. “ Scribner’s Monthly Magazine,” Feb. 1880, p. 
 538; “English Mechanic,” xxxii. 117, 1880; see also Urba* 
 nitzky, “Das Elektrische Licht”, 1883, p. 56. 
 
 + Bohm. See Merling’s “Elektrische Beleuchtung,” p. 
 394, or Urbanitzky, op. cit. p. 63. 
 
26 
 
 has been suggested by Mr. J. T. Bottomley,* 
 who proposes to utilise a Geissler arrange- 
 Anent to exhaust the chamber into which the 
 *foot of the fall-tube of the Sprengel is led, thus 
 putting the two pumps into series. 
 
 Class V.—Injector Pumps. 
 
 There are a few pumps depending for 
 their action upon the principle of the in¬ 
 jector, the degree to which they exhaust 
 depending upon the velocity of efflux of 
 mercury from an orifice, as in the original 
 injector of Hauksbee. The earliest of these 
 were designed by Cavarraf and Plateau.}: 
 Another form, exhibited in 1876 at South 
 Kensington, was invented by Prof, von 
 Feilitzsch,§ in which two cylinders, fitted with 
 pistons, worked by cranks, drove a mercury 
 blast through suitable jets and drew in air, so 
 creating a vacuum. It exhausted down to a pres¬ 
 sure of 1 millimetre, or about 1,300 millionths 
 of an atmosphere. 
 
 Several other injection pumps of the centri¬ 
 fugal species were described by de Romilly|| 
 in 1881, one of them being designated as a 
 fineole. Nothing is known to the writer as to 
 its performance. 
 
 Class VI.—Mechanical Mercurial 
 Pumps. 
 
 Only one pump is known to the writer as 
 coming definitely within this category; and 
 
 this is a pump designed and constructed by 
 Mr. J. Wimshurst, and of which no account 
 has hitherto been published. It consists of an 
 endless chain of little steel buckets, which 
 pass up one barometric column and down 
 another, within steel tubes containing mercury. 
 Below, they enter a mercury bath, where they 
 pass under two square pulleys, rising over a 
 higher driving pulley between the two. The 
 buckets as they descend, mouth-downwards, 
 carry down air from above the top of the 
 barometric column, and discharge themselves 
 as they come up in the mercury bath. Owing 
 to the fact that it has hitherto been found 
 necessary to employ oil as a lubricant, the 
 power of this highly ingenious apparatus to 
 produce a vacuum is limited. 
 
 There are a few pumps concerning which 
 the writer has not been able to obtain infor¬ 
 mation, including those of Diakonoff, Neveux, 
 Pfluger, and Southby, which are known to him 
 by name only. 
 
 Results. 
 
 The results that have so far been obtained 
 by various pumps may be briefly tabulated as 
 follows : the vacua produced being specified 
 both in millimetres and in millionths of one 
 atmosphere. 
 
 If Rood’s method of measurement be correct, 
 the results attained by him are very remark¬ 
 able. 
 
 Authority. 
 
 Nature of Pump, 
 
 Pressure in 
 Millimetres of 
 Mercury. 
 
 Pressure in 
 millionths of 
 one atmosphere. 
 
 Crookes . 
 
 Improved Sprengel (maximum result) . . . 
 
 0*000 046 
 
 1 
 
 TT 
 
 Gimingham. . 
 
 Single-fall Sprengel, ri millim. diam. 
 
 0*000 51 
 0*000 006 
 
 2 
 
 
 Five-fall Spreneel .. . .. 
 
 si 
 
 1 
 
 Rood. 
 
 Plain Sprengel . 
 
 o*ooo 152 
 
 1 
 
 5 
 
 
 Rood’s Sprengel, heated . 
 
 0*000 002 
 
 1 
 
 Bessel-Hagen .... 
 
 Old Geissler, after 25 strokes . ... 
 
 0*110 
 
 145 
 
 if . 
 
 New Geissler (2 taps), after ditto (average) .... 
 
 0*008 5 
 
 II 
 
 > i ...... 
 
 ,, ,, (maximum result) ...... 
 
 0*008 2 
 
 io£ 
 
 
 Old Toepler, after five strokes ... 
 
 0*007 5 
 
 0*006 4 
 
 10 
 
 
 ,, after five more,, . 
 
 8 
 
 
 Modified Toepler (average) ... 
 
 0*000 012 
 
 1 
 
 ^3 
 
 
 ,, „ (maximum result) . 
 
 0*000 008 
 
 1 
 
 * Bottomley. “Rep. Brit. AasoC.,” 1886, Birmingham 
 Meeting, p, 519. 
 
 + Cavarra. “ Comptes Rendils.” 1843. 
 
 + Plateau. Hervorbringung eines Vacuums mittelstderCen- 
 rifugalkraft des Quecksilbers. “Pogg. Ann.” 151. 1843, 
 
 } Yon Feilitzsch. Theorie Und Construction ciner hydro 1 ' 
 
 dynamischen Luftpumpe. Greifswald, 1876. See also 
 “Mitth. des naturwiss. Veretnes v. Ncupommern und Rugen,” 
 ix., 1877; and Catalogue of Loan Collection of Scientific 
 Apparatus (1876), p. 134. 
 
 || F. de Romilly. “ Journal de Physique,” Ser. 1, yol. x., 
 303, 18815 Ser. 2, vol. iv., 366, 1883. 
 
2 7 
 
 Conclusion. 
 
 On comparing the experience of various 
 workers, it seems as if the best class of 
 pump for the production of such vacua as are 
 required for lamps is the third class, as modi¬ 
 fied sp as to drive the air up the pump-head 
 and down a simple short barometric trap into 
 an already partially exhausted chamber. No 
 one appears to have yet tried a shortened 
 Sprengel with a crook in the fall-tube. The 
 writer offers it as a suggestion. Further, if 
 the experiments of Rood are worth anything, 
 they indicate that an immense advantage is 
 gained by working with pumps heated up 
 above the boiling point of water. The “ ad¬ 
 sorption ” of gases and vapours against the 
 surfaces of glass and mercury in the work¬ 
 ing parts of the pump is certainly much 
 less hot than cold. Why should not all 
 pumps be so constructed as to enable this 
 method to be adopted ? Since much seems to 
 depend on the purity of the mercury, why 
 should not the mercury be distilled direct 
 into the pump ? Whenever cements are 
 used, why should not some plastic inorganic 
 substance, such as chloride of lead or 
 tungstate of lead, be employed, instead of 
 resin, pitch, or other organic body, which 
 will give off vapours ? Lastly, if the device of 
 exhausting into an already fairly well ex¬ 
 hausted chamber so greatly improves the 
 degree of rarefaction attainable, why should 
 we not carry this process one or two stages 
 further, and relay a series of pumps one 
 working into the other ? Such a process 
 would resemble those processes of successive 
 operations which have been called “ Pattin- 
 sonisation; ” and it is possible that it might 
 yield results surpassing anything yet attained. 
 
 In surveying the literature of the mercurial 
 air-pump one cannot but be struck with the 
 immense number of workers who have con¬ 
 
 tributed to the invention, and the number of 
 details that have been independently re¬ 
 invented by different individuals. The litera¬ 
 ture of the mercurial pump affords, indeed, a 
 striking proof of the fact that inventions grow 
 rather than are made. The invention is 
 essentially the product of the age in which 
 it appears, a necessary consequence of the 
 
 inventions and discoveries that have preceded 
 
 • 
 
 it. The scientific method of investigating 
 historical events has shown us how false, how 
 childish, is the “ great man” theory of his¬ 
 tory, which was taught—and alas ! is taught 
 still—to us at school. But if the great man 
 theory of history is fallacious, so is also the great 
 man theory of inventions. There were steam- 
 engines before Watt, locomotives before 
 Stephenson, telegraphs before Wheatstone, 
 telephones before Bell, gas-engines before 
 Otto. It may be that occasionally an inventor 
 strikes upon a valuable or useful improvement; 
 it is exceedingly rare for an absolutely original 
 invention to be sufficiently perfect to be of 
 immediate use. Of the essential insufficiency 
 of the great man theory of inventions, the 
 literature of the mercurial air-pump affords a 
 most striking proof. 
 
 The investigation of this literature, which 
 has long occupied the writer, has been a 
 fascinating pursuit, partly because of its 
 unexpected richness, partly on account of the 
 fascination of the subject. Everyone who has 
 worked with mercurial air-pumps must ac¬ 
 knowledge to a kind of fascination in watching 
 the ebb and flow of the liquid metal, and in 
 speculating on the nature of the actions that 
 go on in the vacuous spaces. It was, perhaps, 
 with some such sense that Hauksbee, after 
 describing one of his physico-mechanical 
 experiments, wrote these words :—“ Such a 
 dense and polite Body is Mercury; such a 
 subtile Mover is Air ; and such an apt 
 Repository is an Exhausted Receiver.” 
 
28 
 
 APPENDICES . 
 
 APPENDIX I. 
 
 On Three-way Taps. 
 
 'fhe oldest form of three-way tap—that with 
 a longitudinal boring opening at one or other 
 end of the barrel of the tap—was invented in 
 1675, by Senguerd, Professor of Physics in the 
 University of Leyden, and first described in 
 the second edition (1685) of his “ Philosophia 
 Naturalis.” Such taps—made of metal or glass 
 —were used in most of the earlier pumps 
 described in this paper. The main difficulty 
 in constructing such taps in glass arises from 
 the trouble in fashioning the curved longi¬ 
 tudinal passage. Geissler’s way of blowing 
 and boring such taps is shown in Muller- 
 Pouillet’s “Physik” (1876), i. p. 230. Doubt¬ 
 less this difficulty led to such suggestions as 
 that of Morren, mentioned on p. 8. The most 
 satisfactory mode of constructing such taps in 
 glass appears to be that designed by Weinhold 
 )see p. 9), depicted in Fig. 37. The body of 
 
 Fig. 37. 
 
 Weinhold’s Three-way Tap. 
 
 the tap, T, is simply formed from a piece of 
 hollow glass-tubing, with a short piece of 
 narrower tubing inserted across it, the whole 
 being then ground slightly conical on its 
 
 exterior surface, to fit the conical tubulure. 
 One end of the tube is open, and, as shown in 
 the upper part of the figure, a small hole is 
 pierced in the glass wall at the under side, so 
 that in this position there is a communication 
 from A through this hole to the horizontal tube 
 on the left, through which exhaustion is made. 
 When the tap is turned 90°, to the position 
 indicated in the lower part of the figure, A 
 communicates vertically through the small 
 transverse tube, to the small chamber, M, 
 which is closed above by the simple two-way 
 tap, t 2 , as described on p. 9. 
 
 Babinet’s pattern of three-way tap, alluded 
 to on p. 7, is depicted in Fig. 38. In the 
 
 Fig. 38. 
 
 Babinet’s Three-way Tap. 
 
 first position there is communication vertically 
 through the tap, the horizontal side-tube being 
 closed. By turning through 45°, as shown in 
 the second position, all three apertures are 
 closed. By turning through 45 0 more, so 
 bringing the tap into a position at right angles 
 to its first position, communication is opened 
 between the side tube and the lower tube, the 
 top aperture remaining closed. The taps used 
 by Alvergniat, and by Schuller (Fig. 18), are 
 of this kind. A method of keeping such taps 
 air-tight, by surrounding them with a liquid 
 

 29 
 
 } acket, has been described by Grehant (‘ ‘ Journal 
 de Physique,” ii. 214, 1873). One objection to 
 such taps appears to be their liability to break 
 off transversely in the plane containing the 
 passages. Another objection, which applies 
 equally to all these three-way taps, is the 
 liability to leakage through the lubrication, in 
 which channels form during the working of 
 the tap, as mentioned on p. 7. This evil is 
 lessened, but not entirely obviated, in the form 
 of tap devised by Greiner and Friedrichs, 
 depicted in Fig. 39. Here there are two 
 
 Fig. 39. 
 
 Greiner and Friedrichs’ Tap. 
 
 parallel transverse channels at about 45°, by 
 means of which the lower aperture from the 
 pump-head can be connected at will with 
 either of the two upper openings. The tap 
 has to be turned through 180° instead of 90° 
 as in other three-way taps. The channels in 
 the film of lubricant are, therefore, twice as 
 long, and do not, moreover, lead round directly 
 from one aperture to another. For leakage to 
 occur, in the position shown, from the top 
 left-hand aperture to the top right-hand aper¬ 
 ture, a distance of one whole circumference 
 would have to be travelled round the channels 
 in the lubricant. The risk of leakage is, 
 therefore, reduced to about one-quarter, com¬ 
 pared with other three-way taps. As a 
 lubricant, melted indiarubber is sometimes 
 employed. The writer prefers a soft pomade 
 made of equal parts of paraffin wax and 
 vaseline melted together. Fairley (“ Journal 
 Society Chemical Industry,” 1887) recom¬ 
 mends indiarubber melted into a small 
 quantity of vaseline. 
 
 APPENDIX II. 
 
 On Two-way Taps. 
 
 The ordinary conical glass tap, ground into 
 its socket, and lubricated with stiff grease, is 
 not good enough for working with high vacua. 
 It is liable to leak at the ends; the grease 
 gives off vapours into the vacuum ; it usually 
 
 has too much idle space ; and, like the three- 
 way taps, forms channels in the grease. The 
 two latter points can be mitigated, the one by 
 having very narrow channels, of, say, not only 
 one milimetre bore; the other by having an 
 oblique bore in the plug, as in the Greiner 
 three-way tap. To prevent leakage at the 
 ends it has been usual of late to surround the 
 socket of the tap with an external mercury 
 jacket, or to provide a funnel-shaped collar 
 around the neck of the tap to receive mercury. 
 A convenient form devised by Mr. Eiloart and 
 manufactured by Cetti*, has the socket also 
 closed below and filled with mercury. Quite 
 recently Mr. Eiloart f has described a still 
 better form of air-tight tap, which is illustrated 
 in Fig. 40. The plug of the tap is provided 
 
 Fig. 40. 
 
 with two grooves running round it, which are 
 filled with mercury and check any leakage of 
 air through the socket. To fill the grooves 
 the plug is removed, the socket closed with 
 the end of the finger, filled partially with 
 mercury, and then the plug inserted. To 
 prevent the mercury flowing down the tubes 
 they should be previously closed. 
 
 APPENDIX III. 
 
 On Vacuum Cements. 
 
 For cementing joints, in cases where a 
 fused joint in the glass is not convenient, 
 various recipes have been given. None are 
 equal, however, to a real fused joint. A 
 mixture of resin (clear colophonium) and 
 bees’-wax, in about equal parts, has been 
 employed by Crookes, Moss, Gimingham, and 
 others. It should be applied warm, and the 
 parts to be joined should be well-warmed 
 
 * See “Journ. Chem. Soc.” ccxcvi, 611.] See also 
 Shenstone’s “ The Methods of Glass-blowing ’ (1886), p 66 ) 
 giving instructions how to manufacture this and otner parts 
 of pumps, such as air-traps, &c. 
 
 + Eiloart: “ Chemical News,” lvi. 224,1887. 
 
30 
 
 previously. If there is a greater proportion of 
 resin it becomes brittle. Indiarubber—pre¬ 
 ferably good black unvulcanized gum-rubber 
 —warmed, so as to become sticky, also makes 
 a fair cement. Rood suggested a mixture of 
 96 parts of Burgundy with 4 of gutta-percha. 
 Chappuis (“Wied. Ann,” xii. 167, 1881) 
 suggested a mixture of vaseline and white- 
 wax. The writer has used as a cement a stiff 
 pomade consisting of one part of vaseline with 
 three of paraffin wax. This seems to be pre¬ 
 ferable to organic matters, though probably 
 some mineral cement, such as tungstate of 
 lead or chloride of lead, would be preferable. 
 The fusible material, known to glass-blowers 
 as arsenical glass, or arsenical cement, is 
 preferable in those cases where it can be used. 
 
 APPENDIX IV. 
 
 On the Purification of Mercury. 
 
 So important is the purification of the 
 mercury for use in the vacuum pump that 
 some notice of the various processes is im¬ 
 portant. The methods may be classified 
 under four heads :—• 
 
 I. Mechanical methods. 
 
 II. Chemical methods. 
 
 III. Electrolytic methods. 
 
 IV. Distillation. 
 
 I. Mecha?iical Methods. — Very impure 
 mercury may be roughly cleansed by simple 
 filtration through porous materials, for ex¬ 
 ample, by squeezing it through a dry 
 chamois-leather. Even filtering through a 
 dry paper filter, with a few small apertures 
 pierced in the bottom with a sewing-needle, 
 will suffice to remove films of impurity caused 
 by the presence of lead, zinc, and, their 
 oxides, though no mechanical methods will 
 purify mercury from these metals. Filtration 
 through the pores of wood or cane, by means 
 of the mechanical air-pump, is still more 
 perfect, but it has the disadvantage of 
 partially oxidising the mercury as it falls in 
 spray below the filter. Pure mercury should 
 leave no tail behind it when poured from a 
 glass dish ; and it should not, when shaken 
 up in a dry glass bottle, yield any dark grey 
 powder. 
 
 II. Chemical Methods. — An excellent 
 method is to agitate the mercury in small 
 quantities in a stoppered glass bottle, after 
 having added to it about half its own volume 
 of sulphuric acid (diluted previously with an 
 
 equal amount of water), and a few drops of 
 strong nitric acid. Care must be taken in the 
 process that the stopper is not violently 
 ejected by sudden evolution of fumes. After 
 thus agitating the mercury, it should be 
 thrown into a separating funnel, and drawn 
 off as required. Another method is to agitate 
 it with dilute mercurous nitrate, or with very 
 dilute nitric acid; or it may be boiled with 
 either of these liquids for an hour or more ; or 
 it may be left to digest for several weeks with 
 either of these liquids in the cold in a large 
 shallow dish. Agitation with a solution of 
 ferric chloride has also been recommended, 
 but has the disadvantage of reducing the 
 mercury to a state of fine division, which is 
 re-united only with difficulty. An acid solu¬ 
 tion of bichromate of potash has also been 
 suggested. None of these methods yield 
 mercury anything like as pure as the distilla¬ 
 tion methods. 
 
 III. Electrolytic Methods. —To purify mer¬ 
 cury by electrolysis, a vessel should be pro¬ 
 vided having an internal division extending a 
 few centimetres high, to separate the impure 
 mercury, which serves as anode, from the puri¬ 
 fied mercury, which serves as kathode. Or, in 
 default, a small shallow glass beaker may re¬ 
 ceive the impure mercury (added in small 
 quantities from time to time), and be floated 
 upon the top of the purer mercury in a separa¬ 
 ting funnel. A dilute solution of mercurous 
 nitrate, or of nitric acid, should be added as 
 an electrolyte to such a depth that it covers to 
 a depth of at least one centimetre the partition 
 between the pure and the impure mercury. 
 Two platinum wires must dip into the two por¬ 
 tions of mercury. Two Danielbs cells will be 
 found to be a suitable battery. 
 
 IV. Distillation Methods. —The commer¬ 
 cial process of purifying mercury consists in 
 distilling it under a thick layer of iron filings 
 in retorts of iron, earthenware, or glass. The 
 iron filings are used to prevent spirting. Some¬ 
 times cinnabar is added, in a proportion of 
 about 10 per cent., to prevent the foreign 
 metals present in the mercury, from distilling 
 over with the mercury; they are retained as 
 sulphides. At the ordinary pressure of the 
 air, distillation is however very slow. Distilla¬ 
 tion in vacuo is more rapid, and greatly to be 
 preferred, as the mercury need not be heated 
 to more than 170? or 200° C. Special forms 
 of apparatus for carrying out distillation in 
 vactco have been described by Wild,* Wein- 
 
 * Wild : “ Carl’s Repertorium,’’ vii, 258,1871, 
 
3 T 
 
 hold, * L. Weber, f Pfaundler,^ Clark, § 
 Wright, || and Nebel. 
 
 Most apparatus of this class requires the use 
 of a mercurial pump to start the distillation, 
 which when once started maintains its own 
 vacuum, the drops of distilled mercury falling 
 down a narrow tube as in Sprengel’s pump. 
 This is not, however, the case with Clark’s 
 apparatus, which can be procured from Casella, 
 from F. E. Becker, or from other leading instru¬ 
 ment makers in London. It will distil about 
 800 grammes per hour. The very simple plan 
 described by L. Weber, requires no pump, but 
 is started by filling it with mercury, and in¬ 
 verting it. This apparatus, slightly modified, 
 is depicted in Fig. 41. A piece of glass tube, 
 
 Fig. 41. 
 
 about two centimetres in diameter, and six or 
 seven centimetres long, is sealed obliquely to 
 the top of a barometer-tube, made of glass, of 
 internal diameter not less than four milli¬ 
 metres, and external diameter about 6’5 
 millimetres. Another piece of the same tube, 
 from 20 to 30 centimetres long, is sealed on at 
 a slightly descending angle, and ends in a 
 second barometric shaft, not less than 90 centi¬ 
 metres long, and of 1 *8 to 2 millimetres internal 
 diameter. A glass tap may be inserted in the 
 
 * Weinkold : zb., ix. 69, 1873 ; and xv. i f 1879. 
 
 + L. Weber : zb., xv. 52, 1879. 
 
 t Pfaundler : zb., xv. 328, 1879.; 
 
 ? Clark : “ Phil. Mag'.” [5] xvii. 24 Jan. 1884. 
 
 || Wright: “ Sillim. Journal,” xxii, 479 ; and “ Chcm 
 News,” 1881, 311. 
 
 IT Nebel: “ Exner’s Repertorium,” xxiii. 236,1887. 
 
 bottom of the shorter limb. After filling this 
 apparatus with mercury it is inverted, the ends 
 dipping into two cups, the one, A, containing 
 the mercury to be cleaned, the other, B, to 
 receive the distilled mercury. The level of 
 mercury in A should be carefully watched, 
 otherwise the proper level of the column in the 
 wider tube will not be maintained. A Bunsen 
 burner is applied beneath the widened part of 
 the tube, which should be protected from the 
 direct action of the flame, by being packed 
 round with asbestos, held in a piece of strong 
 wire gauze, which, if properly supported, will 
 serve also to hold up the apparatus. The heat 
 applied at this part volatilises the mercury at 
 the top of the barometric column, and the 
 mercury vapour passes over into the exposed 
 obliquely descending part, which serves as a 
 condenser. The drops collecting here run at 
 intervals into the fall-tube, and in falling main¬ 
 tain the vacuum. This apparatus will not 
 distil so quickly as the more costly apparatus 
 of Clark. 
 
 APPENDIX V. 
 
 On the Vapour-Pressure of Mercury. 
 
 The pressure exerted by mercury vapour at 
 different temperatures has been measured by 
 various physicists, with the following results. 
 The pressures are given in millimetres of 
 mercury. 
 
 * Regnault: “Relation des Experiences,” ii., p. 506. 
 + Bessel-Hagen : “ Wied. Ann,” xvi. 610, 1882. 
 t H. Hertz : “Wied. Ann,” xvii. 199, 1882. 
 
 5 Ramsay and Young: “ Journ, Chcm. Soc.,” 1885. 
 
3 2 
 
 According to McLeod*, who adopted a 
 chemical method of observation, the pressure 
 at 20 0 C. does not exceed 0*00574 millimetres, 
 or about 7*6 millionths of one atmosphere. 
 
 APPENDIX VI. 
 
 Vacuum Gauges. 
 
 The only kinds of gauges considered here 
 are the spark gauge and the McLeod gauge. 
 
 Sftark Gauges.—The ordinary spark gauge 
 consists of a glass bulb, into which are sealed 
 a pair of electrodes, usually plain platinum 
 wires. When a spark or series of sparks are 
 sent across these electrodes from an induction 
 coil, the character of the spark enables the 
 observer to judge of the degree of rarefaction 
 attained. At quite a low degree of exhaus¬ 
 tion, such as is indeed attainable by a good 
 mechanical pump, the spark discharge changes 
 to a pale luminous glow, so well known in 
 Geissler’s tubes. With a higher degree of 
 rarefaction, the “ dark layer” lying between 
 the negative electrode and the luminous space 
 surrounding it makes its appearance ; with a 
 higher degree of exhaustion this dark layer 
 broadens and becomes a dark space. With 
 still higher exhaustion the spark ceases to 
 pass even over so small a length as a quarter 
 of an inch between the wires ; but, as Bessel- 
 Hagen has shown, if metal electrodes of larger 
 area are used, a discharge still takes place at 
 the highest vacua attainable. As spark gauges 
 are very largely used with mercurial pumps in 
 the exhaustion of glow-lamps, it may be 
 remarked that the nature of the metal surface 
 of the electrodes, whether rough or bright, 
 appears to influence the discharge, and to 
 occasion results that are sometimes perplexing 
 and contradictory. An interesting fact pointed 
 out by Bessel-Hagen is that the mere passage 
 of an electric discharge through an exhausted 
 bulb appears to dislodge from the glass walls 
 of the bulb some of the condensed vapours or 
 gases which had settled thereon. Something 
 of the same kind had been previously noted by 
 Gassiot. 
 
 McLeod’s Gauge .— As mentioned above 
 (p. 16) the McLeod gaugef works upon the 
 principle of compressing a known volume of 
 rarefied gas into a smaller known volume, the 
 ratio of the two being accurately known, and 
 then measuring the pressure so applied, and 
 
 * McLeod : “ Chemical News,” xlvii. 251, 1883. 
 
 t McLeod: “ Phil. Mag.,” (4), xlviii. no, 1874, and 
 
 “ Proc. Phys, Soc. Lond.,” i 30, 1874, 
 
 so calculating back to the original pressure. 
 For example, suppose a certain portion of gas 
 occupying the large known volume V under 
 the small unknown pressure to be com¬ 
 pressed under the larger known pressure P, so 
 that it occupies the smaller known volume v , 
 assuming Boyle’s law to hold good, we have 
 
 y>V =2 Pz'; 
 
 whence it follows that— 
 
 v 
 
 i>=P~ 
 
 V 
 
 The resulting number, being the pressure 
 in millimetres of mercury if P is also measured 
 in millimetres of mercury. To convert to 
 fractions of an u atmosphere,” all that is 
 necessary is to divide by 760, because the 
 standard pressure of “ one atmosphere” is 
 taken as equal to that of a column of mercury 
 760 millimetres high. 
 
 McLeod’s gauge is represented in Fig. 42 
 
 Fig. 42. 
 
 in the simple form. V is the bulb whose 
 contents are compressed into the narrow, 
 graduated volume-tube v above it by raising 
 the mercury in a barometric tube beneath. 
 As the mercury rises, it first cuts off the air in 
 V from the rest of the pump, then compresses 
 it, rising simultaneously in the graduated 
 pressure-tube, ^5, through the top of which the 
 guage communicates with the pump. To 
 obviate errors arising from capillarity, v and^ 
 are constructed out of pieces of the same tube. 
 The tubes being graduated in millimetres the 
 pressure P is simply the difference of height 
 between the mercury in the 57-tube and that in 
 the ^5-tube. By multiplying this difference by 
 
33 
 
 the ratio of the capacities of the final and 
 initial volumes, a number is obtained which is 
 approximately the original pressure. To 
 obtain, according to McLeod, a second 
 approximation, the number so obtained should 
 now be added to the observed P, because the 
 height of the mercury in the p -tube is a little 
 lower than it would be if there were no residual 
 
 gas in the pump ; the new number so found 
 
 v 
 
 should again be multiplied by —, giving a still 
 
 V 
 
 closer value for P. A simpler way of getting 
 at the true result is as follows :—When the 
 volume has been compressed to V, the real 
 pressure, P, upon the gas is made up, partly of 
 the observed mercurial pressure, measured as 
 a difference of height, h, in the two tubes, and 
 partly of the residual gas pressure above, 
 namely, p. Hence, in accordance with Boyle’s 
 law, one ought to write 
 
 pV = (/z -j- p) v ; 
 whence p (V— v) = hv\ 
 v 
 
 and p—h -. 
 
 V —v 
 
 This form of gauge has several defects ; one, 
 that the exact amount of air enclosed in V is 
 not well defined, as the mercury, in rushing 
 up the barometric tube, may sweep some air 
 into or out of the mouth of the gauge-bulb. 
 Another defect is, that the range of reading 
 of the gauge is limited. Another, that air- 
 bubbles are apt to creep up into the 
 gauge. Another, that it is difficult to 
 read off accurately volumes at the top 
 of the volume-tube. Modifications have been 
 introduced by Gimingham and by Rood 
 to meet these points. Gimingham provides 
 the bulb with a narrow funnel-shaped aperture 
 below, having a perfectly level orifice (see 
 Fig. 43), so that the mercury on rising cuts off 
 a perfectly definite volume of air. He further 
 provides the gauge with a volume-tube consist¬ 
 ing of two parts, a wide part, v lf for measuring 
 the larger residua, and above it a very narrow 
 part, v 2 > f° r measuring the smaller residua of 
 
 air or gas. Two pressure-tubes, p t and p 2 are 
 used, one made of the wide tubing, the other 
 of the narrow tubing; the mercury rising in 
 both at the same time. Gimingham uses two 
 Crookes’s air-traps in the barometric column 
 below the gauge, as shown in Fig. 23 (p. 17) 
 at t t. Rood does not fuse up the top of the 
 volume-tube, but closes it in a manner that 
 leaves the interior cavity cylindrical, by insert¬ 
 ing and cementing in with Burgundy pitch and 
 gutta-percha, a small closely-fitting glass rod, 
 the lower end of which has been ground off 
 horizontal. Rood also corrects the volume- 
 
 Fig. 43. 
 
 A 
 
 Gimingham’s Modified McLeod’s Gauge, 
 
 reading by allowing for the curvature of the 
 meniscus at the top of the mercury, making 
 special preliminary experiments for that pur¬ 
 pose. He also raises and lowers the mercury 
 in the barometric column by pushing up a sur¬ 
 rounding glass tube, closed at the bottom, 
 partially filled with mercury. This can be 
 raised to any desired height, and its lower end 
 supported on a pile of blocks of wood, and 
 adjusted by inserting thinner pieces or pieces 
 of cardboard. Toepler’s and Bessel-Hagen’s 
 method of using the Toepler pump as its own 
 gauge is described on page 21 above. 
 
34 
 
 DISCUSSION . 
 
 The Chairman said he had listened with great 
 pleasure to this most interesting paper; before he 
 came into the room he thought that he knew some¬ 
 thing about the mercurial air-pump, but his only 
 feeling 'now was how much there was still to bo 
 known. His name had been mentioned in connec¬ 
 tion with high vacua, but he must say that his idea of 
 a vacuum varied very much as the work went on. In 
 his early days, when they worked by mechanical 
 means alone, an exhaustion of a millimetre was 
 looked upon as very good indeed, and a little beyond 
 that would have been thought almost a perfect 
 vacuum. Then Sprengel’s pump was introduced, 
 and with this we easily got the barometer gauge 
 and the barometer itself level, and that was then 
 called a perfect vacuum. Then the apparatus was 
 used for the various physical experiments which 
 enabled them to detect the pressure of residual air, 
 and the vacuum was in this way seen to be very 
 imperfect. Then chemical means were employed, 
 such as filling the tube with carbonic acid and 
 absorbing by potash. Then those taps which figured 
 in nearly all the pumps had to be done away with, 
 for no good could be- obtained if there was a tap or a 
 lubricant. When all this was effected, as in the best 
 form of Sprengel, the vacuum suddenly improved, 
 until it would not conduct electricity. Finally, by 
 McLeod’s gauge, with the modification of Giming- 
 ham, we could actually measure the vacuum, and 
 then we found we had what was scarcely worth 
 calling a vacuum at all. The best he had ever 
 succeeded in getting was the one hundredth of a mil¬ 
 lionth of an atmosphere, which was equivalent to 
 one-tenth of an inch at the top of a barometer tube 
 200 miles in height. That sounded like a very good 
 vacuum, but when a small tube containing a centi¬ 
 metre of air was exhausted to that extent, there 
 would still be left in it ten billion molecules. The 
 nearer they got to perfection, the farther off it 
 seemed, and the more hopeless it appeared to think 
 of getting a perfect vacuum by any available means. 
 
 Mr. J. Swinburne, referring to the relative advan¬ 
 tages of the long and short Sprengel pump, said he had 
 tried a good many of each, and he must say he pre¬ 
 ferred the long one. The short one looked better on 
 the table, but the advantage was more apparent than 
 real. If any air got taken down a certain part of the 
 shaft, say six inches, it was safe enough, and never 
 got back again if the pump worked properly. The 
 chief disadvantage to the short shaft arrangement, 
 was the difficulty of preserving a mechanical 
 vacuum. Those he had to do with in commer¬ 
 
 cial work he had found very troublesome, and 
 he would never use one in lamp manufacturing 
 if he could help it. He preferred to use a long shaft 
 with air pressure to lift the mercury. He should 
 like to know the Chairman’s opinion as to the 
 measuring of these vacua, because the figures given 
 seemed to him purely fancy figures, and reminded 
 him of the candle-power of arc lamps. In the first 
 place, the McLeod gauge must have some mercurial 
 vapour in it ; he did not know whether the tension 
 of that had been accurately determined, but, accord¬ 
 ing to Regnault, it was about 5 ° millionths, and if 
 that was so, it was absurd to talk of getting a 
 vacuum anything below that. The next point which 
 was put was the way in which air, gas, or vapour 
 seemed to flatten out against the glass. You might 
 have what seemed to be a good vacuum; when 
 mercury was let into it slowly, it would show a 
 bubble, but if admitted quickly would go right up 
 with a smack, and there would be nothing visible at 
 all. In opening bulbs or lamps under mercury, con¬ 
 siderable care must be taken, and then many would 
 fill completely with mercury, which would mean a 
 perfect vacuum. Of course, if the McLeod gauge 
 was worked like that, it showed there was some¬ 
 thing deceptive. Again, taking several lamps of! 
 the same pump, and testing them, it would be 
 found that different results were obtained both with 
 coils and opening them under mercury. A good 
 Geissler pump was its own gauge. Referring to the 
 drawings of his own pump, which had been described 
 by Professor Thompson, in which there was a bend, 
 Q, and a little chamber, m, with a valve at the 
 top, he said he had never used it commercially, 
 for it was a far better pump than was needed 
 for that purpose. The object of the bend and 
 the chamber was to meet the point which had been 
 urged; all the pumps shown exhausted into a 
 moderately good vacuum, but that little pump ex¬ 
 hausted into a vacuum which was as perfect as any 
 of the other pumps could produce. The vacuum in 
 M was as perfect as it was in A in the other pumps. 
 By exhausting into the little chamber M, and 
 taking, say, 20 strokes, you could get a vacuum 
 to as many millionths as you felt inclined, but 
 about the accuracy of all such measurements he 
 was sceptical. In the first place, if it could be 
 depended on, the vacuum would increase in geo¬ 
 metrical progression, but if you worked to any 
 point you liked, and then left it for two hours 
 and came back, you found quite a different state 
 of things. Mercury vapour gave very different 
 effects when tested with a coil and in other 
 
35 
 
 ways. In such a pump as that, it seemed impervious 
 to a spark until it was put in connection with a lamp, 
 and then immediately there was a green phos- 
 phoresence under ordinary circumstances. That 
 seemed to show that mercury vapour was a non¬ 
 conductor, unless some air was mixed with it. He 
 believed that had been discovered at the beginning 
 of the century. He had always followed the practice 
 of distilling the mercury, because when it was once 
 put into one of these pumps it lasted until the pump 
 was broken. Silicate of soda was a good cement, and 
 got perfectly hard. He did not know whether he had 
 mentioned, in the paper he sent to the Electrician , 
 that he had tried the same system of elevation as 
 Nicol, except that the three shafts were all joined to¬ 
 gether in one piece, instead of being joined with 
 india-rubber. But there was a great difficulty in 
 regulating the flow of air, and if by any means it 
 went wrong, the mercury got into the phosphoric 
 acid tubes, and into the lamps. As to the relative 
 merits of the Sprengel and the Geissler, the general 
 idea was that the Geissler did to rough out with, 
 and the Sprengel to finish the exhaustion. He 
 should prefer to use a Sprengel for roughing out, if it 
 would do it, but he always finished with a Geissler. 
 He used the form shown on the diagram, with the 
 siphon bend and second chamber, and in making 
 comparative tests he had always found that the 
 Sprengel stopped working long before the Geissler. 
 
 Professor W. Ramsay said Professor Thompson 
 deserved the greatest credit for having brought 
 together sound valuable information. With regard 
 to the McLeod gauge, he thought it possible that 
 some of the air really flattened itself against the glass, 
 or, perhaps more correctly, adhered to it, and formed 
 a sort of condensed gaseous film, so that the 
 ultimate amount measured was not the actual 
 amount in the bulb, but only a small fraction of it. 
 It was quite certain that if the temperature were not 
 kept quite even, a correct measurement could not be 
 made. As a practical point in connection with the 
 cleaning of Sprengel pumps, he would recommend 
 the use of glacial acetic acid, which had such a high 
 vapour pressure that it removed itself. Even with 
 the cleanest mercury one got films of oxide formed, 
 and in the Sprengel pump, where the little drops fell, 
 there was always a film of oxide formed. This was 
 readily dissolved with glacial acetic acid, and another 
 advantage connected with its use was that you need 
 not pour it in, you only need attach a vessel 
 containing it to the apparatus, and the vapour would 
 pass over in sufficient quantity to clean the pump. 
 
 Mr. R. P. Sellon said there was one point worth 
 notice, which Mr. Swinburne had already touched 
 upon. When you tested the vacuum of a glow-lamp by 
 opening a globe under mercury, you sometimes got a 
 very different result from that which you obtained by 
 testing with a Ruhmkorff coil. That was probably due 
 
 to the fact that the residual air in the globe, broken 
 under mercury, became condensed, and flattened out 
 against the sides of the globe, and thus escaped 
 
 detection. 
 
 • 
 
 Mr. C. V. Boys said he had not done much with 
 mercurial air-pumps, but he remembered when Prof. 
 Rood’s form was described making one, and testing 
 it with a McLeod gauge. The special points about 
 it were the heating, the crook, and another point 
 which had not been referred to. As to the crook, 
 there was no doubt that a small quantity of air 
 always stuck round the upper end of the column, 
 that was swung over and caught on the lower side of 
 the crook, and he was satisfied that was a good 
 thing. The effect of heating was most remarkable ; 
 he kept the apparatus so hot that he could not touch 
 it, and the first visible effect was that the reflection of 
 the mercury in the tube was much brighter, as if it 
 were more intimately in contact with the glass; in fact, 
 the film of air which undoubtedly sometimes existed 
 wasto a great extent squeezedout. The otherpointwas 
 this: that the mercury, according to Professor Rood’s 
 instructions, was not fed into the top of the pump, 
 as in all others of the Sprengel type, but was allowed 
 to fall in a small stream through a bulb. When it 
 began to work, the bulb was entirely filled with 
 mercury, and the equator of the bulb was beneath 
 the level of the fall-pipe. As soon as the vacuum 
 began to get pretty good in the fall-tube, the mer¬ 
 cury in the tube fell, and there was above it a very 
 good and very hot vacuum. Then the mercury, in 
 an exceedingly fine stream, was delivered through 
 that vacuum hot, and gave up there any entangled 
 air and moisture, and fell into the space below in as 
 good a condition as mercury could be. He tried one 
 of these pumps with a McLeod gauge. In a hot 
 pump, when there was no crook, the hammering noise 
 was enormously greater. Of course, this hammering, 
 owing to the vibration which ensued, allowed a 
 minute quantity of the air which was carried down 
 to escape back again. The introduction of the 
 crook stopped the hammering, and, of course, pre¬ 
 vented the escape of air. There was no reason why 
 people should not always use distilled mercury. 
 Clark’s apparatus was very simple, and you could 
 at a push, distil as much as 14 lbs. per day with it, 
 which would probably be sufficient for all commercial 
 purposes. He was sorry Professor Thompson had 
 not said more about the centrifugal form of pump. 
 About ten years ago he had an idea that it would 
 work very well, and set about making one, 6 in. high 
 and 6 in. in diameter. What corresponded to the 
 fall-tube in a Sprengel pump were radial arms with 
 the ends curved in, and he was in hopes it would 
 work well, but the result was it did not work at 
 all; he did not get more than about a foot of vacuum 
 altogether. He believed the reason was the shape of 
 the radial tubes, and that if, instead of being straight 
 they had been curved like a turbine, the pump would 
 have worked. However, he did not care to undertake 
 
making such a thing in glass, and let the matter drop. 
 He had heard considerable doubts expressed whether 
 the figures of the McLeod gauge represented what 
 was supposed. As to the flattening of a minute trace 
 of residual gas between the mercury and the glass, 
 although that will happen when the mercury was let 
 into a bulb, he fancied that under the circumstances 
 in which it was used in the gauge, the mercury 
 slowly r ising, with a large expanse above it, would 
 allow the residual gases to rise until they were caught 
 in the small tube, and then there would be such a 
 small surface of mercury that there would not be 
 much room for error on this account. 
 
 Mr. W. M. Mordey said he felt much indebted to 
 Professor Thompson for pointing out so clearly where 
 to go for information on this matter. With regard 
 to paraffin wax as a cement, he might say that about 
 nine years ago he made a barometer of rather thick 
 glass, and happened to break the end in sealing it. 
 He simply dipped it in paraffin wax and filled it up 
 in the usual way, and it had worked quite well ever 
 since. He quite agreed with Professor Thompson’s 
 theory as to invention. He believed an inventor was 
 one who, in ninety-nine cases out of the hundred, gave 
 audible voice to whispers that were in the air. 
 
 Mr. Boys said he had found a mixture of pitch 
 and gutta-percha make an excellent cement; but of 
 course no organic compound would compare with 
 sealed glass. 
 
 Professor Thompson, in reply, said Professor Rood 
 gave a recipe for cement, consisting of 96 parts of 
 Burgundy pitch and 4 of gutta-percha. There was a 
 very convenient form of mercury distiller now avail¬ 
 able, much simpler than Clark’s, and one which any 
 amateur could easily make with two pieces of baro¬ 
 meter tube ;’it was described by Weber, and he would 
 put a rough sketch of it on the board. Sulphuric 
 acid would remove itself as well as glacial acetic acid, 
 thougn not perhaps so quickly. He had hoped that 
 Prof. Ramsay could have given from his own re¬ 
 searches some more precise figures than Regnault 
 as to the vapour pressure of mercury. Regnault gave 
 it at about 20 millionths of an atmosphere at the 
 freezing point. 
 
 Professor Ramsay said that should be divided by 
 about 200. 
 
 Professor Thompson said Bessel-Hagen gave it as 
 26 millionths at 20° C., and H. Hertz at about 
 17 millionths. In the figures Bessel-Hagen gave, he 
 distinctly stated that he subtracted from the observed 
 result in the gauge (which was simply the top 
 of a Toepler pump) 26 millionths as the sup¬ 
 posed pressure of mercury vapour, and assumed 
 that the remainder represented the pressure of 
 the residual air. That answered the point which 
 
 Mr. Swinburne had raised. Bessel-Hagen made 
 exactly the same remark as Mr. Swinburne, 
 as to the Geissler pump going on extracting the air 
 longer than the Sprengel. There was no doubt that 
 air pressing upon mercury or glass did adhere to the 
 surface; that was one reason why, exhausting from a 
 chamber in which there ’already had been made a 
 moderately good vacuum, you got a much better 
 result. The mercury was then returned into 
 the pump without having had any great air- 
 pressure put upon it. For that reason he differed 
 from Mr. Swinburne as to the use of air-pressure to 
 lift the mercury. No doubt he was correct in saying 
 it was difficult to get workmen to make a satis¬ 
 factory mechanical apparatus ; plumbers did not 
 know very well how to make tubes air-tight, 
 and to make them vacuum-tight was still more 
 difficult, but he did not see why a mechanical 
 difference of this kind should be allowed to stand in 
 the way if the pump would work better when the 
 mercury was never exposed in any part of its journey 
 to the ordinary pressure of the atmosphere, but only 
 worked from a moderately good vacuum to a more 
 perfect one. For this reason he had emphasised the 
 peculiarities of the later forms of pump, which ex¬ 
 hausted only between one stage of vacuum and 
 another. 
 
 The Chairman, in proposing a vote of thanks to 
 Professor Thompson, said McLeod’s gauge was no 
 doubt imperfect, but it had been found that by taking 
 certain precautions the indications could be relied on, 
 within a certain margin, very well indeed. For 
 instance, the mercury must be squirted into the 
 pump through as good a vacuum as the pump 
 had up to that point reached, and the surrouuding 
 vacuum would take all the air from the surface of 
 the mercury, so that it entered the pump freed 
 from air-films. The pump and everything con¬ 
 nected with it must be heated to as high a point 
 as was safe for some time before trying to get a 
 measure of these high vacua. It was true the measure¬ 
 ment was only that of the residual air, not of 
 mercury vapour, but that was easily managed. 
 The mercury vapour diffused very slowly through 
 long spiral tubes, and it could easily be kept out by 
 a little device which he published some time ago. 
 By putting between the bulb to be exhausted and the 
 pump a long tube containing in the middle of it a 
 little iodine, and on each side of it powdered sulphur, 
 the mercury would not pass the iodine because it was 
 converted into a solid iodide of mercury. The iodine 
 would not pass the sulphur because it was converted 
 into iodide of sulphur. But the sulphur would, 
 perhaps, get into the pump, and so they put 
 powdered silver outside the sulphur to keep 
 the sulphur out. In that way you might keep 
 the whole mercury vapour out, and the gauge 
 would only show the residual air which was in 
 it. It struck him some time ago what a blessing it 
 would be to physicists if gallium were cheaper; it was 
 
37 
 
 a metal which was liquid at 86 9 Fahr., and would 
 remain liquid at temperatures much below that ; it 
 gave off no vapour, and did not oxidise in the air, so 
 that it was an almost perfect metal for a pump of this 
 kind. There was an alloy of gallium and aluminium, 
 which was much lighter, but unfortunately it decom¬ 
 posed in water. There was one point in connec¬ 
 tion with, this paper which specially commended 
 itself to the Society, and that was that with¬ 
 out these mercurial pumps we should have had 
 no incandescent electric lighting. Another point 
 was that it had given a new employment to women. 
 Formerly glass-blowing was entirely confined to men, 
 and when he wanted pumps made commercially at 
 first he had to pay^io or ^15 a-piece for them. In 
 
 connection with some others he started a factory 
 in which the whole glass-blowing was done by 
 girls. It was easy work, just suited for their 
 delicate fingers, and they became as skilful as any 
 of the men. The wages were good, the hours 
 short, and a great deal of work could be done 
 at home. A Sprengel pump, for which he used to 
 give ^10 or ^15, the girls thought they were hand¬ 
 somely paid for making at 10s. He believed most 
 lamp manufacturers now employed girls to do the 
 glass blowing, for the pumps were always breaking, 
 and had to be repaired on the spot. 
 
 The vote of thanks was carried unanimously, and 
 the meeting adjourned. 
 
 PRINTED BY W, TROUNCE, 10, GOUGH-SQUARE, FLEET-STREET, E.C 
 
. 
 
 
 
 
 
 
 
 
 
 
 
 
 *■ 
 
 
 
 
 
 • I 
 
 
 
 
 
 
 
 
 
 
 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 I 
 
 

 
 
 
 
 ■ A 
 
 
 
 ■'* . *: 
 
 
 
 ■ * 
 
 . 
 
 V 
 
 
 
 
 
 
 , 
 
 
 
 
 
 : 
 
 
 
 
 
 . 
 
 
 
 
 
 
 • 
 
 
 
 • 
 
 
 
 
 
 
 
 
 
 
 
 
 . ■ 
 
 
 
 
 
 . 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 . 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 ' 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 '