TS 320 .S4 Copy 1 DEPARTMENT OF COMMERCE Scientific Papers OF THE Bureau of Standards S. W. STRATTON, Director No. 396 THERMAL AND PHYSICAL CHANGES ACCOMPANY- ING THE HEATING OF HARDENED CARBON STEELS BY HOWARD SCOTT, Assistant Physicist H. GRETCHEN MOVIUS, Assistant Physicist Bureau of Standards SEPTEMBER 20, 1920 PRICE, S CENTS Sold only by the Superintendent of Documents, Government Printing Office,', Washington, D. C. Washington government printing office 1920 DEPARTMENT OF COMMERCE Scientific Papers OF THE Bureau of Standards S. W. STRATTON, Director No. 396 thermal and physical changes accompany- ing the heating of hardened carbon steels BY HOWARD SCOTT, Assistant Physicist H. GRETCHEN MOVIUS, Assistant Physicist Bureau of Standards SEPTEMBER 20, 1920 PRICE, S CENTS Sold only by the Superintendent of Documents, Government Printing Office, Washington, D. C. WASHINGTON GOVERNMENT PRINTING OFFICE 1920 tp Digitized by the Internet Archive in 2011 with fidrftfirftf from OCT 29 1929, The Library of Congress http://www.archive.orig/details/thermalphysicalcOOscot ^ THERMAL AND PHYSICAL CHANGES ACCOMPANY- ING THE HEATING OF HARDENED CARBON STEELS By Howard Scott and H. Gretchen Movius CONTENTS Page I. Introduction 537 II. Experimental method 539 III. Heat evolution Ac, 540 1 . Effect of rate of heating 544 2. Effect of tempering temperature 545 3 . Effect of time at tempering temperature 546 4 . Effect of composition 548 5. Effect of austenitic structure 549 IV. Relation of changes in physical properties to heat evolution 551 1 . Martensitic steel 551 2. Austenitic steel 554 V. Summary 555 I. INTRODUCTION The widespread interest which has been recently expressed in the properties of steel in the "blue-heat" range and in the subject of "temper brittleness" makes it highly desirable to study in detail the transformations in steel below the Aj change. In a previous paper 1 the authors have pointed out certain thermal characteristics of the magnetic change in cementite as observed in annealed steels by means of thermal analysis. In this paper the subject under investigation is the thermal change observed in hardened steels on heating below Ac t . Outside of the possible bearing of such information on the low- temperature properties mentioned, there remains the desirability of establishing fundamental characteristics of steel. The one in question is of particular value in that it may furnish a practical basis for defining the natural boundary between martensite and the troostite of tempering, which from present information is very indefinite. A survey of the changes in some of the physical properties of carbon steels on tempering would, on account of certain incon- sistencies, lead one to doubt the existence of a sharp demarcation between the constituents — martensite and troostite. Heating 1 Chemical and Metallurgical Engineering, 22, p. 1069; June 9, 1920. 537 538 Scientific Papers of the Bureau of Standards Ivoi. 16 curves of hardened steels, however, have shown a well-marked heat evolution ending around 300 C. Such heat evolution would be expected from the usual conception of the formation of mar- tensite; namely, that one or more of the transformations occurring on slow cooling are suppressed by quenching. The consummation of the suppressed transformation (or transformations) is a mani- festation of the completion of the constitutional change and, therefore, evidence of a boundary between two constituents, pre- sumably martensite and troostite. Whether the end of this heat evolution should be used to define those constituents the future will decide ; the present work seeks only to establish its nature in a variety of carbon steels and its relation to accompanying changes in some of the physical properties. In the literature some work has appeared on this heat evolution in hardened carbon steels. Osmond 2 and Maurer 2 have given inverse-rate heating curves; Heyn and Bauer 3 and Portevin 3 have given differential curves showing the phenomenon. The temperature values for the transformation are somewhat higher than those obtained here. In general, the curve inflections are neither prominent enough nor the statement of operating details sufficient to allow of a precise definition of the transforma- tion characteristics. Also the effect of important variables has not been determined. This phenomenon has been observed also by continuous measurement of the changes in some physical properties on heating. Grenet * detected an inflection in the ex- pansion and electric -resistance curves of a high-carbon steel, Chevenard 5 in expansion curves, and Honda 6 in magnetic- induction curves. The magnetic curves are the only ones which seem to follow closely the progress of the heat change. Brush 7 has made extensive observations on the heat evolution at ordinary temperatures in recently hardened steels. He noted a heat evo- lution, greatest immediately after hardening, gradually diminishing in rate with time and becoming imperceptible after several weeks. The physical changes accompanying this spontaneous evolution were very small in comparison with those accompanying even slight tempering. While the present research is confined to carbon steels because of their fundamental importance, it is being extended to alloy 2 Osmond, J. Iron and Steel Inst., p. 38; No. 1, 1890. Maurer, Rev. de Met., 5, p. 711; 1908. 3 Heyn and Bauer, J. Iron and Steel Inst., 79, p. 109; 1909. Portevin, Rev. de Met.. 13, p. 9; 1916. * Grenet, Rev. de Met., 1, p. 353; 1904. 6 Chevenard, Rev. de Met., 14, p. 610; 1917. a Honda, Sci. Reports Tohoku Imp. Univ., 6, p. 149; 1917. 7 Brush, Bull. A. I. M. M. E. No. 153, p. 2389; 1919. Scot! "1 MoTiusj Thermal Changes of Hardened Steels 539 steels in order to obtain further light on the effect and function of the alloying elements. II. EXPERIMENTAL METHOD The inverse-rate method of obtaining thermal curves has been used at the Bureau of Standards as the most effective and satis- factory method for studying the transformations in steel. Used in connection with the apparatus already described 8 excellent curves can be obtained at the low temperatures where the heat evolution under examination is found. The details of mounting, size of sample, and operation are given in the above reference. A temperature interval corresponding to 20 microvolts on a platinum, platinum-rhodium thermocouple was used in this investigation. TABLE 1. — Results of Chemical Analyses of Steels Investigated c Mn Si P S Per cent 0.40 .44 a. 46 a. 73 .95 1.01 1.94 Per cent Per cent 0.01 .02 ,06 .01 .24 .01 .01 Per cent Per cent 1.00 .35 .38 .22 0.02 .04 .02 0.05 .05 .01 .005 .005 a Furnished by courtesy of Carnegie Steel Co. The materials studied were seven steels of the compositions given in Table 1. Heating for quenching, as noted on the curves and in the tabulated results, was carried out on the prepared samples by introduction into an electrically heated alundum-tube furnace wound with resistance wire. Charcoal was present to reduce oxi- dation, and a platinum, platinum-rhodium thermocouple was used for the measurement of temperature. In tempering, the specimen was heated 30 minutes in an oil or nitrate bath, as required by the temperature. Temperatures below 300 C were measured with a mercury thermometer. All samples of the quenched 0.95 per cent C steel not receiving subsequent tempering were run within from 1 to 3 days after quenching. All the other steels not tempered were run within from 6 to 16 days after treatment, excepting the 1.01 per cent C and 1 .94 per cent C steels (curve j) , which were run the day follow- ing treatment. 8 Scott and Freeman, Bull. A. I. M. M. E. No. 152, p. 1429; 1919. Also B. S. Sci. Papers, No. 348. 54-0 Scientific Papers of the Bureau of Standards III. HEAT EVOLUTION Ac t {.Vol. 16 The principal phenomenon under consideration here, the heat evolution on heating hardened carbon steels, will be designated as "Ac t ," a notation used by one of the authors • for the same phe- nomenon in a high-alloy steel. °C 500 WO 300 200 /oo \ 0.75 Per cent C Stecf Quenched from 800 °C \ !© ® © I© '}-£" ■-E {-/V =;'-/V -At :■■-£ ■:.-// K <3-c> \-J3 \ Time interval in seconds /O IS 10 IS IS 20 20 2S ¥5 SO SS FlG. I. — Inverse-rate heating curves of hardened steel, showing effect of rate of heating on Act The thermal curves, taken to show the effect of several variables on the transformation Ac t , are shown in Figs. 1, 2, 3, 4, and 5. For the reduction of the thermal-curve data to tabular form, the • Scott, Bull. A. I. M. M. E. No. 146, p. 157; Feb.. 1919. Also B. S. Sci. Papers, No. 335. Scett 1 Meviusl Thermal Changes of Hardened Steels 54i Off 5 Per cenf C Stee/ X Quenched /n . E\ -,£- r -£• | : j | >— VT 5 ' : 1 Cm \ \ \ \ i --S v \ \ -s \ I \ 1 i ■ } f i \ 1 j i i I ZOO /OO 1 I \ I -B i 1 1 i i 1 1 \—JB \ "3 i \ \ i. 5 : \ \ i ! '; "l \ \ Time Jnferisa/ /r? ■seconds /s zo zo 20 /s J_l • zo /s /s 1 1 1 zo zo 1 1 1 /S /S ZO 1 1 1 Fig. 2. — Inverse-rate heating curves of hardened steel, showing effect of previous tempering for jo minutes on Act 542 Scientific Papers of the Bureau of Standards Wot. r« temperatures of the principal curve bends caused by the Ac t trans- formation were taken as denoted on the curves by B, M, and E, beginning, maximum, and end, respectively. The rate of heating given is that just before the beginning of the transformation. The values in the column of Table 2 marked "Intensity" represent SOO WO 3O0 2O0 /oo 0.95 Percent C Steel Oaenchec/ /r> o/7 from Time interna/ in secone/s /s so /s so /s so /s so /s so /s so Ftg. 3. — Inverse-rate of heating curves on hardened steels, showing effect of duration of previous tempering on Act the difference in seconds between the time at the maximum and at the end of Ac t , except in the case of the austenitic steel, where they represent temperature drop. The temperature of the maximum of A^ (maximum temperature before decalescence when that phenomenon was observed) is also ScoU "1 Mtrvius} Thermal Changes of Hardened Steels 543 Carbon Steels Quenched from above /?c a °c C .03 Mn - Si .0/ MO Mb MU .73 .95 I.OI — .35 1.00 .38 .ZZ — .01 .Ob .OZ .01 .21 .01 4,00 || 11-11 i 5 ! Mi! i . '( I I ! i j SOO i ! ! . f t 1 j i • 1 , III' l i •! } i i '■ 1 i ! 1 • ! - s \ \ \ \ ¥00 t '■1 I'll , \ \ \ \ \ i ! i ■: •. i i \ 300 ! \ \ 111 ! i i '. i | '. \ -, \ i t 1 s )-£ )~ •-" .-••'"/••■""" ;.•.— ! {-M \ \ \ \ e i \ \ \ \ \ \ \ \ V ■ \ \ \ ■■■ \ \ \ \ i ZOO JOO \ \ ; i \ i 11 \ \ \ i i_ ■:_ '— \ \ \ \ ■••. \ \ I i i 7//77e interval in seconds /5 ZO ZO ZO ZS ZO ZO £0 II 1 1 1 1 1 1 1 1 1 1 1 II es 30 1 1 Fig. 4. — Inverse-rate heating curves of quenched steel, showing effect of composition 4218°— 20 2 544 Scientific Papers of the Bureau of Standards Ivoi. 16 given, but the heating curves are not plotted to show Aci, in order to avoid excessive reduction of the curves on reproduction. 1. EFFECT OF RATE OF HEATING Rate of heating has a considerable effect on the temperature and form of Ac t for the comparatively fast rates required by ther- "C WO 500 WO 300 £00 too r^u^tenific Quenched in wafer Iron- from Carbon /7//oy ;/oo°c S ame *W ed in liquid air \ \ t i > ;■ \ \ > K i i J © if. . I. : © ® 1 © i I, > V \ — At \~M 1 J/tf' -"-=-^ — >-• • 130" -frj/r s -zrj° £85° ' \ 1 \--B :■ \ 1 Time 15 20 25 JO Ji 1 1 1 1 1 1 1 interval in s 5 /O /5 ZO 45 Mill econds 5 10 1 1 1 IS zo 1 1 1 IS 20 25 30 OS MINI Fig. 5. — Inverse-rate heating curves of austeniiic iron-carbon alloy {1.Q4 P er cent O mal analysis, as may be seen from the curves of Fig. i , which were taken on the 0.95 per cent C steel. The principal data taken from these curves are plotted with rate of heating as the abscissas, in Fig. 6. It will be noted from this figure that the temperature characteristics of Ac t for zero rate are 155, 250, and 260 C, respec- tively, for the beginning, maximum, and end of the transforma- tion. This appears to represent the progress of the transformation Scott 1 Mvvius} Thermal Changes of Hardened Steels 545 for a tempering time approximating normal tempering conditions, probably about 30 minutes. From the sharpness of the begin- ning of the transformation it would appear that the quenched steel is the equivalent of a steel instantaneously cooled and then drawn in the neighborhood of 150 C. Fig. 6 illustrates this interesting point: That for the size of specimen used in this case there is no appreciable difference between the thermal characteristics of an oil-quenched and of a water-quenched specimen. j 2. EFFECT OF TEMPERING TEMPERATURE The heating curves represent the progress of tempering for a necessarily very short time at any temperature in the Ac t range. To show the effect of hold- ing for a definite time at several tempering temper- atures on the characteris- tics of Ac t , heating curves were taken on specimens of the 0.95 per cent C steel quenched in oil from 8oo° C and tempered 30 min- utes at the temperatures given in Fig. 2 and Table 2. From a consideration of these data it may be seen: (1) That the beginning of Ac t is from 10 to 17 C higher than the tempering temperature when that is above 200 C ; (2) that the transformation is com- pleted at a temperature between 250 and 270 C; and (3) that for each tem- perature up to 250 C there is a definite and character- istic form of curve. The estimated temperature of the end of Ac t for zero rate (260 C) is, therefore, from (2) in practical agreement with the end of the transformatior for a tempering period of 30 minutes. From (1) and (3) it is evident that the heating curves might be used to esti- mate the previous tempering temperature within certain limits. Rate of heating .30 X/sec. Fig. 6. — Effect of rate of heating on temperature and intensity of heat evolution of 0.95 per cent C steel 546 Scientific Papers of the Bureau of Standards ivoi. 16 It may be of interest to note that, for a tempering temperature of 270 and 300 C (Fig. 2) , there is a slight deflection of the curves to the right, indicating an absorption of heat over the range of about 350 to 450 C, which is in conformity with the observations of Heyn and Bauer 10 under similar conditions. 3. EFFECT OF TIME AT TEMPERING TEMPERATURE It has long been recognized that the time of holding at a tem- pering temperature has a very considerable effect on the resulting physical properties, and it is even held that a long time at a low temperature is equivalent to a short time at a higher temperature. The thermal curves of steels tempered for different lengths of time in the Ac, range should, therefore, throw some light on the validity of this much discussed proposition. In Fig. 3 heating curves are given to show the effect of main- taining a steel for different lengths of time at the tempering tem- perature. The 0.95 per cent C steel, hardened by being quenched in oil from 8oo° C, was used. Specimens were maintained for 5, 30, and 60 minutes at each of the two tempering temperatures, 200 and 230 C, chosen because they represent temperatures at which tempering is well in progress, but not to such an extent as to eliminate the thermal effect. It may be noted from these curves and the compiled data of Table 2: (1) That the beginning of Ac t is higher for a long than for a short exposure at the tempering temperature; (2) that the intensity of the transformation is less for a long tempering period than for a shorter one; and (3) that the rate of progress of the transformation is greater at the higher tempering temperature than at the lower one. From (1) and ^(2) it is apparent that time has a decided effect on the transformation characteristics. The third conclusion is evident from the fact that at 200 C an exposure of 60 minutes is necessary to reduce markedly the intensity of Ac t , while at a tem- perature only 30 C higher the intensity is much more strongly reduced by a 30-minute exposure. This is in agreement with the tempering experiments of Barus and Strouhal, 11 whose measure- ments of electrical resistance and thermal emf show the rate of transformation to be much greater at the higher tempering tem- peratures in the Ac t range. This indicates further that the tem- pering time of 30 minutes used in the preceding section represents actual equilibrium or zero-rate conditions at the temperature of the end of Ac t , though, of course, not at lower temperatures. 10 See footnote 3. "Barus and Strouhal. Bull. U. S. Geological Survey. No. 14; 1885. Scott "I Moviusi Thermal Changes of Hardened Steels TABLE 2. — Thermal Characteristics of Hardened Carbon Steels 547 Composition Mn Si Heat treatment Quench- ing temper- ature Quench' ing me- dium Tem- pering tem- pera- ture Time at tem- pering tem- pera- ture Rate of heat ing Act temperature Be- gin- ning Maxi- mum End Intensity ot heat evolution Aci max- imum Per cent Per cent Per cent .95 .24 .40 .46 .44 .73 .95 1.01 .35 1.00 .38 .22 .01 .06 .02 .01 .24 .01 800 800 800 800 (") 800 800 800 800 800 800 800 800 800 800 800 800 800 800 900 900 900 900 1,100 1,100 1,100 1,100 Water. ..do... ..do. Oil... Oil... ..do.. ..do. ..do. ..do. ..do.. ..do. ..do. ..do.. Oil... ..do. ..do. ..do. ..do. ..do.. Water. ..do... Oil-'... Water. Oil.... Water. Water. ..do... ..do... ..do... 200 230 250 270 300 350 400 200 200 200 230 230 230 Min- utes °C /sec. Seconds (1) Effect of heating at different rates 0.23 183 290 319 .16 178 285 308 .05 162 261 282 .22 167 273 295 .23 7 14 8.5 733 736 (2) Effect of tempering at different temperatures 0.12 167 273 295 .13 215 275 300 .13 240 282 307 .15 267 310 .16 .15 .12 .17 . 15 8.5 9.0 1. 5 0.5 733 733 733 734 734 733 733 733 735 (3) Effect of tempering for different periods of time 0.15 204 276 300 .13 215 275 300 .16 224 273 300 .14 224 271 296 .13 240 282 307 .13 244 289 350 9 9 4 5.5 1.5 0.5 (4) Results for different compositions 0.12 167 259 278 .12 174 270 286 .13 174 276 297 .12 175 271 292 .12 167 273 295 .10 167 261 284 731 731 727 731 733 728 (5) Results for austenitic structure .08 174 308 319 .11 182 347 361 .15 184 353 366 .12 179 302 317 ' C Temp, drop 2 24 731 a Annealed. 6 Air cooled. 548 Scientific Papers of the Bureau of Standards {Vol. 16 It may be inferred from the nearly identical characteristics of Ac t following tempering for 60 minutes at 200 C and 5 minutes at a temperature 30 C higher, that these two treatments produce the same structural condition, but, because the rate of transforma- tion changes with temperature, it does not follow that this par- ticular relation holds quantitatively for any other temperatures in the Ac t range. In general, however, the effect of time may be regarded as equivalent to that of temperature within limits as far as the characteristics of Ac t are a criterion of the constitutional changes in the steel. 4. EFFECT OF COMPOSITION For the sake of comparison of the several martensitic steels investigated, the temperature values of Ac t taken from Table 2 have been given a small correction on the basis of Fig. 6 to reduce them to a constant rate of heating of o.io°C per second; these values are given in Table 3. By comparing the synthetic steels in the first group, low in manganese, or the commercial steels in the second group, containing 0.20 to 0.40 per cent manganese, with respect to the variable carbon, one may see that the maxi- mum and end of Ac t are somewhat higher for the higher carbon contents and that the transformation intensity is approximately proportional to the carbon content. TABLE 3. — Transformation Characteristics of Martensitic Steels for Rate of Heating of 0.10° C per Second SYNTHETIC STEELS C Mn SI Ac, temperature Intensity Beginning Maximum End Per cent 0.40 1.01 CarSd/7 Too/ S/ee/ Q.95; /Vs?,.ZZ;S/,.ZV " ~°*% Quenc/ied m water from 300 "C ov. Sa/7 • Sam i/b/e r7 " b/e £ . \/i>yZby /6"> /00 ZOO 300 WO 500 7~err7fier/ny temperature (,00 700 FlG. 8. — Effect of tempering temperature on scleroscope hardness of o.Q$ per cent C steel that the change is about 85 per cent complete at 300 C, or in the Ac t range. The density curve 5 is somewhat irregular, but the maximum rate of change occurs in the vicinity of the end of Ac<. From the foregoing analysis it is evident that the changes in the physical properties considered are related very closely to the heat evolution Ac t , particularly in the case of the magnetic proper- ties, maximum induction, and coercive force. These relations are forceful indications of a natural boundary between martensite and the troostite produced at about 260 C on tempering. Such a boundary should be detectable also by the changes in micro- Scott "I Movnts} Thermal Changes of Hardened Steels 553 structure. Authorities, however, differ on the temperature of this boundary for simple steels, and place it anywhere in the range from 250 C to 400° C. Careful observers have studied this change, and while not suggesting an end point, have made observations indicative of one in the region of the end of Ac t . Howe and Levy, 16 after quenching eutectoid carbon steel from 1100 C to water, find that on a 5-minute exposure to 300 C 700 600 \ 1 *soo X Quenched in wafer from <900 " C /OO 200 J00 400 500 600 700 "C Tempering temperature Fig. 9. — Effect of tempering temperature on Brinell hardness of 0.Q5 per cent C steel the original white martensite needles are almost completely broken up. Heyn 17 notes a coarsening of the needle structure at 275° C. These observations are indicative of a structural change in the vicinity of the end of the heat evolution. The region under investigation is thus narrowed down, and future observers should have little difficulty in denning precisely the nature of the accom- panying changes. 16 Howe and Levy, Trans. A. S. T. M.. 16, part II, p. 7; 1916. " Heyn quoted by Sauveur, The Metallography and Heat Treatment of Steel, p. 304. 554 Scientific Papers of the Bureau of Standards IVol. 16 2. AUSTENITIC STEEL In a foregoing section (p. 550) attention was called to a sharp change in direction of the heating curves of the austenitic steel. This inflection, denoting an abrupt increase in the rate of heat evolution, was noted to start at about the temperature at which /00 200 J00 VOO 500 Te/nfoer/na temperature too 700 'C Fig. 10. — Change of physical properties with tempering tem- perature of martensitic carbon steels the heat evolution in a martensitic steel begins to disappear. It may therefore be of some interest to compare this thermal behavior of the steel with the density changes in similar steel. In Fig. 1 1 are plotted density values given by Maurer 18 for a 1.66 per cent C steel quenched in water from 1,050° C. The resulting structure is not completely austenitic, but nearly so. u See footnote 2. Scott "| Moviusl Thermal Changes 0} Hardened Steels 555 The curve shows three distinct regions: (1) 20 to 150 C, in which the density increases as in a martensitic steel; (2) 150 to 250 C, in which a drop occurs which recalls the second stage of the heat change of the austenitic steel; and (3) above 250 C, in which it follows the normal course of a martensitic steel. Since martensitization implies a decrease in density (see the black circles of Fig. 1 1 , representing the density change on immersion in liquid air), the second step with a density drop is evidently attributable to completion of the change from austenite to mar- tensite, which is more or less transformed into troostite. The G/.66; A/r?,.09; S/, ./O Quenc/iet? from /050"C (flanrer) 'cm 3 V.75 \7.70 7.05 -ZOO +200 VOO 600 800 'C Te/nfoer/hg femjberafure Fig. 11. — Change in density with tempering temperature of semiaustenitic carbon steel (Maurer) augmentation of the heat evolution of the austenitic steels, as previously explained, is therefore definitely verified. V. SUMMARY The transformation, observed as an evolution of heat, on heating curves of hardened steel has been designated here as Ac t , and its characteristics as revealed in carbon steels have been investigated. The effect of several variables was noted with the following conclusions: 1. An increase in the rate of heating raises markedly the tem- perature of Ac t for a 0.95 per cent C martensitic steel and has a yet more marked effect for an austenitic carbon steel. For zero rate of heating there appears, however, to be little, if any, difference between the principal temperatures, whether the steel is of high or low carbon content or whether it is martensitic or austenitic. The principal temperatures for the 0.95 per cent C martensitic 556 Scientific Papers of the Bureau of Standards Wot. 16 steel were found to be 155, 250, and 260 C, respectively, for the beginning, maximum, and end. 2. The results obtained for specimens tempered at different temperatures before taking heating curves confirm substantially the temperature of the end of Ac t just given. 3. Tempering for a short time at a temperature within the Ac t range has an effect on the transformation characteristics similar to tempering for a longer time at a somewhat lower temperature. 4. The heat evolution of the austenitic steel takes place in two steps, the second being probably connected with the transition from austenite to martensite. 5. A survey of the changes in some physical properties of mar- tensitic carbon steels through the tempering range leads to the conclusion that these changes are all directly related to the heat evolution observed, but only in the case of the magnetic proper- ties, coercive force, and maximum induction is the change of the same type. 6. The change in density of a semiaustenitic carbon steel pro- ceeds in steps similar to the heat evolution of the austenitic steel. 7. The changes in microstructure on tempering martensitic steels are unquestionably related to the heat evolution, but further study is necessary to establish fully this relation. The end point (260 C for zero rate) of Ac t may very properly be taken as the natural boundary between martensite and the troostite of tempering, representing as it does the end of the transformation suppressed on rapid cooling. The competent assistance of H. A. Wadsworth has greatly facilitated this investigation. Washington, April 22, 1920. LIBRARY OF CONGRESS 003 129 912 3