THE PROPEETIES OF AMMONIUi3.l XQmggTE. PABT I. 1887 CXXXL-The Properties of Ammonium Nitrate. Part I. The Freezing Point and Transitim-temperatures. By REGINALD GEORGE EARLY and THOMAS MARTIN LOWRY. AMMONIUM nitrate is remarkable amongst salts both for its ewy fusibility and for the fact that the salt exists in not fewer than five crystalline forms the ranges of stability of which are marked by a series of four well-defined transition-temperatures. These different forms are conveniently named in the same way as the various forms of iron by Greek letters starting with the form which is stable a t the lowest temperatures. The crystalline forms of these modifications and the published data in reference t o the transition-temperatures are set out in the following table.TABLE r. Changes of State in Ammonium Nitrate. Form of Crystalline Change 1 t toliq. Liquid 127" 124-125" 124" 123*5-125.8', 1 6 to E { 125.6" 122.6" 126.2" 1244-L-125*6O 125.0' 125.2". 82*5-86" 81" 82*5-86.5" ) yt,o 6 {8722*S0 83.0" 86-" 85.S0 84" 84O, salt. system. of state. Transition temperature. 165-106° 165" 168O 1652O. -t Cubic 6 "Rhombohedra1 82-26". } B to (36" 31-35' 35O 31-35' 32.4" y TRhombic 1~ Rhombic I 32*2" 35*0° 35.0" 32.5". a Tetragonal a to - 16" - 18". * Or tetragonal. j- Or monosymmetric. I n view of the fundamental character of these constants and of the fact that the temperatures are all easily accessible a redeter-mination was made which has had the effect of fixing the freeain 1388 EARLY AND LOWRY THE PROPERTIES OF point with an estimated error of a few tenths of a degree at 169*6O, whilst the three upper transition points have been determined probably within a few hundredths of a degree at 125'2O 8 4 .2 O and 32*1°. I. Freezing Point at 169.6O. The difficulty of determining the correct melting point or freez-ing point of the salt and the fact that values much below the correct figure have so often been recorded arise from two charac-teristic properties of the salt namely the very great sensitiveness of the freezing point to the influence of small quantities of moisture (details in reference to which will be given in a later paper) and the tendency of the salt to retain its moisture even when attempts are made to remove it by somewhat drastic methods. On account of these properties of the salt we obtained in our earlier experiments a long series of concordant values for the freezing point which were afterwards proved to be more than 20 too lbw.These freezing points were determined by observing the arrest of temperature which occurred immediately after crystal-lisation had begun in about 40 grams of the molten salt the salt being stirred vigorously in a glass tube surrounded by an air-jacket in a bath of boiling water in order to prevent over-rapid cooling. The temperatures were measured by a thermometer graduated in fifths of a degree from Oo to 200° and recently re standardised t o 0 . 0 2 O a t the National Physical Laboratory ; this therrnometar was of the compensated type and the zero had remained constant within Om020 over a period of about ten years.Two auxiliary thermometers were used to record the temperatures of the " exposed column " of mercury inside and toutside the glass tube containing the salt. A series of freezing points determined in this way gave very concordant figures the corrected values being as follows: 166*95O 166*95O 166.90° 167*15O 167*0°. Mean 167.0°. This was the freezing point of a sample of the salt which had been recrystallised twice from water and dried first in a steam-oven during several hours and then during a day and a night over sulphuric acid in a desiccator exhausted with the help of a Gaede mercury pump. Doubt was first thrown on the correctness of these readings when it was discovered that a commercial sample of ammonium nitrate which had been dried and ground and then re-dried in the usual way in the laboratory gave a freezing point of 168*8O that is, nearly two degrees higher than our best laboratory specimen of the salt.As this higher freezing point could obviausly not be du AMMONIUM NITRATE. PART I. 1389 to impurities which would tend to lower the freezing point of the salt it aould only be attributed to the fact that the drying had been accompanied by a grinding which was absent in the labora-tory process of purification already described. In order to secure equally favourable conditions in the labora-tory the purified salt was heated in a steam-oven cooled and crushed in a mortar and left overnight in a vacuum desiccator, this cycle of operations being repeated three or four times.A series of freezing points of material prepared in this way gave the following figures : 169*16O 169*14* 169*04O 169-05O 169.06O. Mean 169.09O. The first two values were for ammonium nitrate from Hopkin and Williams twice recrystallised; the next two are for the same sample after storing for some weeks in a desiccator over calcium chloride; the last value is for a commercial sample of Norwegian nitrate recrystallised from water and dried by the method just described. These experiments showed that the freezing point of the salt was a t least 169.1° and might be as high as 169.2O in view of the first two values recorded in the list. A further improvement in the freezing point was effected by sealing up in an exhausted flask connected with a bulb of phos-phoric oxide some of the salt prepared as described above.The nitrate was shaken up repeatedly to expose a fresh surface and a t the end of a week the freezing point was determined. I n plotting the cooling curves for these determinations of the freezing point, two modifications were made in the practice previously adopted, namely (i) the outer water-jacket was kept a t 80° instead of looo in order to reduce the amount of water vapour round the apparatus; (ii) the stirring of the molten salt was stopped as soon as the temperature showed signs of becoming steady in order to reduce the loss of heat from the1 molten mass and so to prolong as much as p i b l e the horizontal arrest in the cooling curve. The freezing points determined in this way were 169.58O 169.55O.After three months' further drying by phosphoric oxide no further rise of the freezing point was produced but on the contrary it appeared to have receded slightly. The freezing point of the pure dry salt may therefore be placed a t 169.6". If the freezing point of the salt is expressed in whole numbers the figure 170° is more exact than any hitherto given since the experimental numbers recorded above are unlikely to be too high but may still be slightly below the true maximum figure. The drying of ammonium nitrate which is not really a difficul operation when dealing with the aolid salt cannot be effected merely by stronger heating since the salt which begins to decom-pose at 200° retains even a t this temperature the moisture pro-duced by decomposition.Thus two samples which had been heated to vigorous effervescence a t 225O,'froze a t 166.8O and 167*0° when cooled again to the freezing point of the molten salt. Again, a sample of the nitrate which melted originally a t 169O froze a t 1 6 5 ~ 5 ~ after being heated to 230° to 240° for a few minutes and at 163*4O after a second heating. The molten salt indeed appears to cling to its water quite as tenaciously as concentrakd sulphuric aaid and it is only by grinding the solid to a fine powder that complete drying is readily effected. 11. Transition-t emperature at 125.2O. When the molten salt is allowed to cool from the melting point in a tube provided with a thermometer dipping into the salt the highest of the transition-temperatures referred to in table I gives rise only to a retardation of cooling a t about 1 2 5 O .I n order to produce slower cooling and to secure if possible a horizontal arrest in the cooling curve in spite of the poor thermal conductivity of the salt the following method was adopted. About 90 grams of the purified salt were melted and poured into a small beaker which was lowered into the centre of a vacuum-jacketed cylinder A standard thermometer was inserted and the salt was allowed to set round the bulb. The stem of the thermometer passed through a large rubber stopper which served to close the top of the cylinder and 80 prevent loss of heat by convection to the outer air. The cylinder in its turn was immersed as far as possible in a bath of hot water stirred mechanically and maintained a t a fairly constant temperature by a flame.A Bunsen valve was provided for the escape of air by expansion from the cylinder whilst two auxiliary thermometers gave the temperatures of the " exposed column " of the standard thermometer. An attempt was made to retard the cooling still further by exhausting the air from the interior of the cylinder but this procedure did not appear to serve any useful purpose and was abandoned because of the additional trouble caused by the frotxng up of the salt to a porous mass during exhaustion. When observing the transition point a t 125O the temperature of the outer water-bath was maintained a t 95O but it was then generally allowed t o cool to 7 5 O in order to observe the further transition a t 84O to which reference is made in Section I11 below.The readings obtained in this way with air in the vacuum cylinder were 125*Z0 125*6O 125*8* 125.1" 125'6" 125*2O whils AMMONIUM NITRATE. PART I. 1301 with the air exhausted from the cylinder the readings were 125.2O and 125.40. ThO mean of the whole series is 125’4O. Zawidzki (Zeitsch. physiknl. Chern. 1903 47 721) obtained a mean value of 125*0° f o r arrests in the cooling of ammonium nitrate alone and mixed with silver nitrate. I n the case of the transition point a t 84O the poor conductivity of the salt may be compensated by stirring the powder in a small revolving drum heated externally by means of a vapour-jacket but this method cannot be employed a t the higher temperatures as the powder begins to cake into tough lumps.A much more effective method of securing good conduction during heating and cooling consists in stirring the powdered salt in a bath of liquid not differ-ing too widely from i t in density. The liquid selected for this purpose was a mixture of tribromoethane (D 2-62; b. p. 1 8 8 O ) and xylene (D 0.86; b. p. 140°) the density of ammonium nitrate a t this temperature being about 1-6. Good results were also obtained by using nitrobenzene (D 1.22; b. p. 211O). This method of improving the thermal conductivity has the advantage that the thermometric measurements can be made within a maximum of accuracy. The procedure was as follows. The liquid was made into a (‘mush” with powdered ammonium nitrate in a glass tube 20 cm. long and 3.7 cm. in diameter provided with a stirrer and a split cork to carry a standard thermometer; as before two auxiliary thermometers were used to record the temperature of the The glass tube was surrounded by a metal cylinder 20 cm.long and 5 cm. in diameter closed at the top with cotton wool and forming a narrow air-jacket round the tube. This cylinder was supported by a bung in the axis of a larger metal cylinder 32 cm. long and 10 cm. in diameter con-taining paraffin of high boiling point to serve as a heating or cool-ing bath and provided also with a thermometer. I n plotting a heating or cooling curve the paraffin bath was adjusted by hand t o a temperature a few degrees above or below the transition point. The ((mush” in the inner tube was stirred a t intervals of thirty seconds to two minutes according to the velocity of heating or cooling and the thermometer was read ‘immediately after.The arrest-temperatures observed in this way during heating and cooling together with the conditions under which the heating and cooling were carried out are shown in table 11. exposed column ” of mercury 1392 EARLY AND LOWRY THE PROPERTIES OF TABLE 11. ArTest-teirLperutzcres of Anz?nonium iVitl.de Suspe~Zed in a bath of Liquid. (corrected). outer bath. Composition of liquid. Arrest-points Temperature of Heating 126.21 130" Nitrobenzene. 128.16 140 Xylene and ethylene bromide. 125.38 140 Xylene and ethylene bromide. 125.22 130 Xylene and tribromoethane. 125.25 130 Nitrobenzene. Mean 126.24" Cooling 126-27" 120" Xylene and ethylene bromide.126.21 120 Xylene and ethylone bromide. 125.09 120 Xylene and tribromoethane. Mean 125.19' These values may be compared with those givea by Vogt (Physi-I d . Zeitsch. 1911 12 ll29) who obtained with a dilatometer the upper and lower limits 125.25O and 125*13O mean 125-2O. The eight values now recorded range from 125'09O to 125.38O'. The mean of the five readings obtained by heating the salt is 125*24O whilst the mean of the three readings obtained by cooling is 125-19O. The general mean of all the readings is 125.22O and this is probably the best value for the transition-temperature. I n view however of the difficulty of reading the temperatures to O*0lo we prefer to give the transition-temperature to a tenth of a degree at 125.2". 111. Transition-temperature at 8 4 .2 O . The transition a t 84O is accomplished by an abrupt expansion which frequently made itself manifest by fracturing the glass vessel in which a cast sample of nitrate was being cooled. There is how-ever a very strong tendency for over-cooling to occur and in many cases the salt was cooled to 32O without any indication that this change of crystalline form had taken place. . I n order to ensure the conversion of the 6- into the y-form i t was necessary to inocu-late the surface of the block with particles of the salt which had been heated to 60° and to assist the conversion by scratching the surface with a sharply pointed glass rod. Even under these con-ditions the over-cooling was usually very pronounced ; the change of crystalline state (even in the vacuum-jacketed apparatus used successfully to record the change of state a t 125O) only began when the temperature had fallen to 82O or below and the latent heat was then not sufficient t o restore the temperature to the transition point or to produce a horizontal arrest in the cooling curve.I AMMONIUM NITRATE. PART I. 1393 every case therefore the transition merely produced a sinuous curve the highest temperature recorded for the maximum on this curve being 8 2 - 7 O . The first well-marked arrest of temperature a t the transition point was observed when heating a sample of the powdered salt in a machine in which the nitrate was thrown over and over in such a way that the bulb of a thermometer was constantly bathed in the falling nitrate.A small machine in which this principle was embodied gave well-defined arrests on heating a t 84*3O 84*Z0 84'3O, and 8 3 ' 3 O the first three values being concordant within one-tenth of a degree. It is remarkable that this transition-temperature, which was found to be so exceptionally difficult to locate by the methods first employed proved to be by far the easiest of the transition points to determine exactly when once the proper con-ditions were established. The most f avourable conditions for determining this transition point were those already described in connexion with the change of state a t 125*2O namely to compensate fpr the lack of thermal conductivity by stirring the powdered salt in a bath of liquid of almost equal density. The liquid used for this series of experi-ments was a mixture of ethylene bromide (D 2.18; b.p. 131O) and xylene (D 0.86; b. p. 140O). Special precautions were again needed to prevent " lag " in the change of state; the '' mush" was therefore inoculated during heating with crystals heated to looo, and during cooling with crystals heated to 60°. The inoculation was carried out just before reaching the transition point and was followed by gentle stirring. The arrest points recorded in this way which are lower than the mean of Zawidzki's arrests a t 85'4O, but higher than the dilatometric readings 82.16O to 82.36O of Vogt, are set out in the following table: TABLE 111. Arrest-temperatures of Ammonium Nitrate suspended in a Bath of Liquid. (corrected). outer bath. Composition of liquid.84.16 90 Y 9 9 9 84.25 90 9 9 Y Y 84-20 90 9 9 9 9 Arrest-points Temperature of Heating 84.20" 90" Xylene and ethylene bromide. Mean 84.20' Cooling 84.28" Falling slowly. Xylene and ethylene bromide. 84.13 9 9 7 9 9 9 9 9 84-18 9 7 9 9 9 9 9 9 Mean 84.20 1804 EA.RLY AND LOWRY THE PROPERTIES OF In this case the means of the four values obtained during heat-ing and of the three values obtained during cooling are identical. The transition point may therefore be fixed probably within a few hundredths of a degree a t 84*20° or may be given to a tenth of a degree as 84.2". I V . TransitiolL-temperature at 32-10. (a) Heating and Cooling Curves.-The transition-temperature in the neighbourhood of 32O is more easily observed than either of those occurring a t a higher temperature since a prolonged arrest in the neighbourhood of 3 2 O can always be detected when the nitrate is cooled through this temperature; indeed both this arrest and that a t 8 4 O are constantly encountered when the nitrate is handled commercially.I n spite of this fact exceptional difficulty is experienced in securing an exact determination of this transition-temperature. This difficulty was ultimately traced to the fact that over a range of about a quarter of a degree on either side of the true transition point the velocity of change of state is so slight as to be practically imperceptible with the result that the con-version usually takes place a t a temperature definitely below the transition point on cooling and a t a temperature definitely above it on heating.Large numbers of cooling curves were plotted in order to deter-mine this transition-temperature accurately. I n some experiments only a sinuous cooling curve was obtained but in others well-marked horizontal arrests were recorded a t the following tempera-tures : Cast blocks ............... 31*9O 32'0° 32*0°. Pressed blocks ............ 31*7O 31*5O 31*4O 3 1 ~ 8 ~ . Loose crystals ............ 31-6O. Horizontal arrests were also1 recorded sometimes a t lower temperatures for example 31.0° 29-5O 29-6O. Arrests during heating were always a t a higher temperature and the curves wen3 generally of a sinuous form rising to a maximum value before falling again to a minimum approximating to the transition-temperature of the salt.These minima in the heating curves were observed a t the following temperatures : Cast blocks ........................ 33.2O. Pressed blocks ..................... 33*8O 33*1° 33'2O. In one experiment in which the sample had become very much over-heated before the change of state set in a still higher read-ing was obtained a t 35*3O (compare Zawidzki Zoc. cit. who obtaine AMMONIUM NITRATE. PART I. 1398 tan average of 35-0° for seven arrests in heating ammonium nitrate alm0 and mixed with silver nitrate). I n view of the fact that the horizontal arrests or maxima in the cooling curves were always very much more fully developed than the minima in the heating curves it was believed that the former could be assumed to give a correct value for the transition-tempera-ture which was located provisionally a t 32O; actually however the data now quoted can only be used to prove that the transition-temperature lies within certain limits for example between 32-0° and 33*1° and i t was not possible to secure an absolute deter-mination by this method in view of the fact that in no case were the arrests on heating and cooling within one degree of each other.Attempts to secure more accurate readings of these transition-temperatures by using larger quantities of nitrate up to a kilo-gram resulted in failure the conductivity of the salt being so low as to prevent the effective flow of heat from one part of the mms to another; better results were indeed always obtained by heat,-ing or cooling much smaller quantities of ammonium nitrate, insulated as carefully as possible for example in a vacuum vessel, in order to reduce the rate of heating or cooling t o a minimum.A distinct improvement in the heating curves was however, obtained as in the case of the transition a t 84O by stirring about 700 grams of powdered nitrate in a small drum surrounded by a steam-jacket in such a way as to produce a constant flow of nitrate past the thermometer. By using this method the following arrests in the heating curves were recorded: 32.6O 32.75O 32.7O 32*9O 33*0°. These readings are definitely lower than those recorded previously when cast or pressed blocks of the nitrate were heated but no improvement could be effected in the cooling curves so that there still remained a gap of about 0*6O between the highest arrest on cooling and the lowest arrest on heating, The method of stirring the powdered nitrate in a bath of liquid of equal density which had proved so successful a t the higher temperatures was a complete failure when applied to the deter-mination of the transition-temperature at 32O.The temperature recorded on the thermometer immersed in the liquid frequently failed to show any arrest a t all. Even after inoculation the cool-ing curves were extremely erratic and only on two occasions were satisfactory arrests observed a t 31-8O and a t 31.6O; on no occasion was an arrest oherved in the heating curve. It may be noted, however that Muller (Zeitsch. physikd. Chem. 1899 31 354) obtained a satisfactory arrest a t 32*20 by cooling from 80° a mix 1396 EARLY AND LOWRY THE PROPERTIES OF ture of 100 grams of ammonium nitrate with 15 to 20 grams of water and that this temperature lies within 0*lo of our final value for this transition point.(b) Experiments with the BiZatometer.-The unexpected difficul-ties which were encountered in trying to determine the exact posi-tion of the transition-temperature a t 3 2 - 1 O can be traced to the relative slowness with which the change of state takes place immedi-ately above or below the transition point. even when assisted by inoculation as contrasted with the much greater velocity of the changes a t 84O and 1 2 5 O . This is in accordance with the general rule that changes of this character become more and more sluggish as the temperature falls, by reason of the decreasing mobility of the molecules and the increasing resistance which the rigidity of the material opposes to molecular rearrangement.Under such conditions the thermal method becomes difficult or impossible and i t is usually necessary to fall back on some method of determining the transition-tempera-ture in which ample time can be allowed for the change of state to reveal itself. I n the case of ammonium nitrate the most promising method was to follow by means of a dilatometer,* the expansion or contraction which accompanies the change of state, instead of relying on the absorption or liberation of latent heat to produce an arrest in the heating or cooling of the salt. Experiments which were made 'on these lines gave us our first trustworthy value for this transition-temperature and also pro-vided valuable information as to the velocity of the change in the irrmediate neighboarhood of this point'.The solid used in the dilatometer was made by fusing pure dry ammonium nitrate pouring into a mortar breaking the cast lump into pieces about 0.3 cm. in diameter and sieving to free it from dust. This form of the salt was used in order to secure good thermal contact between the solid and liquid and a t the same time t o avoid the risk of fracturing the bulb by the sudden expansion of a closely-packed powder. The dilatometer held about 60 c.c. and the bulb was sealed off after filling about three-quarters full with fragments of nitrate. The liquid was a paraffin of high boiling point which had been treated with concentrated sulphuric acid to free i t from olefines and then dried over metallic sodium; it was introduced in an air-free condition by making use of the apparatus shown in Fig.1 of a paper by Wade and Merriman (T. 1912 101 2430). I n a dilatometer filled in this way there are a t the transition-* Compare vctn't Hoff Zeitsch. ph?jsikat. Chem. 1895 17 130 and '' Vorlesungen,' 1898 vol. i. p. 18 AMMONIUM NITRATE PART I. 1397 temperature two alternative positions fox the meniscus ; one which may be called TTB is the position when all the nitrate is still in the stable low temperature or &form; the second which may be called Vy is the position to which the menisous rises when all the nitrate has been converted (without change of temperature) into the lighter y-form.The position 17s can be determined by heat+ ing the dilatometer from the atmospheric temperature to a point just below the transition point; Try can be determined by heating the dilatometer say to 50° until the whole of the nitrate has passed into the y-form and then cooling it to a point immediately above the transition point at 32O; very little extrapolation is then required to give the exact position of these two points. I n using this method to determine the transition-temperature of ammonium nitrate it is esseiitial that the dilatometer should contain both forms of the salt in order that change of state may take place quite readily in either direction. This condition was secured by heating the dilatometer t o 50° when the meniscus rose to a point well above Vy; on immersing the dilatometer in cold water part of the salt reverted from the y- to the &form a change which was revealed a t once by the appearance of white patches on the lumps of nitrate; this change took place long before the contents of the bulb as a whole had time to cool to the transition-temperature.Having made sure in this way that the nitrate in the bulb contained a substantial proportion both of the 8- and of the y-forms of the salt it was possible to find a range of tempera-tures (just above the transition point) a t which the meniscus tended to settle down in the neighbourhood of Vy by reason of the complete conversion of the contents to the y-form and a range of temperatures (just below the transition point) a t which the meniscus would settle down in the neighbourhood of V S owing to the complete conversion of the nitrate to the &form.The actual behaviour of the dilatometer when heated a t different temperatures after the preliminary treatment described in the preceding paragraph was as follows : (i) When the dilatometer was immersed in a thermostat a t 32*2O the meniscus settled down very quickly to a definite posi-tion between VB and Vy and during the course of two hours showed no tendency to rise to Vy or to fall towards VB. The change of state appeared in fact to be arrested a t the point to which it had been brought by the more drastic preliminary treat-ment of the salt. A precisely similar behaviour was observed a t 32*1° 32.0° 31*9O 31*8O and 31-7O.(ii) When the temperature of the thermostat was reduced to 31.6O the meniscus for the first time began to show a tendency t 1308 EARLY AND LOWRY THE PROPERTIES OF fall towards VB although the change a t this temperature was so slow as to be almost imperceptible; on repeating the experiment with the thermostat set to 31*5* however a definite movement towards VCrp could be seen. These two temperatures are therefore definitely below the transition point of the nitrate. (iii) When on the other hand the temperature of the thermostat was raised to 32.3O a definite but exceedingly slow upward movement of the meniscus towards Vy was observed and a more pronounced movement when the temperature was raised to 3 2 ~ 7 ~ . These experiments on account of the extreme sluggishness of the change of state in the immediate neighbourhood of the transi-tion point failed to fix the exact position of this temperature, although they served to locate it between 31-6* and 32.3O.These two limits agree quite closely with those arrived a t from a study of the cooling and heating curves which had shown arrests below and above tw40 corresponding limits a t 32*0° and 32.6O; rather clmer limits were recorded by Vogt. who observed equal and o p p site slow changes of volume a t 32*40° and 32.62O. (G) Quantitatiue Experiments with the Di1atometer.-In order to determine the exact position of the transition point a series of quantitative experiments was made on the velocity of the change of state as shown by observations with the dilatometer a t tempera-tures above and below the transition-temperature.The preliminary treatment of the nitrate was much the same as in the previous experiments and was carried out as follows. The two extreme positions of the meniscus a t the transition-temperature were first located on the scale of the dilatometer as follows : V/3=20 cm.; Vvy=48 cm. Starting with cold nitrate in the 8-form the dilatometer was next immersed in a bath of water a t 34O in order to initiate the change from the r& to the y-form. A t this temperature (on account of thermal expansion) the two extreme positions of the meniscus would be about 22 and 50 cni.; when therefore the meniscus had risen to about 36 cm. on the scale of the dilatometer it was clear that the two forms were present in roughly equal quantities whilst the thermal conditions of the dilatometer and its contents had not been aeriously disturbed.After this preliminary treatment the dilatometer was immersed in a thermostat set to a definite temperature just above or below the transition point. Thermal equilibrium was quickly estab-lished and exact measurements could be made of the rate of movement of the meniscus consequent on the change of state fro AMMONIUM NITRATE PART I. 1380 to y. or vice versa; the uniformity of this movement O V ~ T con-siderable periods of time was such as to justify the view thafi the rates recorded in heavy type in table IV are definite physical con-stants of the change of state over the range of temperatures from 3 1 . 2 O to 33-0°. The readings represent the movement in mm.of the meniscus during successive intervals of ten minutes. The mean steady velocities (obtained by averaging the numbers which are shown in heavy type in table IV) are recorded in similar unite. 65 50 45 40 35 30 25 20 15 10 5 31 The final column shows the percentage of the salt which would undergo change of state in one hour as deduced from the volume changes recorded in the previous columns. The thermometer readings are given to tenths of a degree but are all subject to a correction of +0*03O. It is remarkable that a t temperatures a whole degree above or below the transition point nearly an hour is required to change half of the material from the B- to the y-form, or vice versa. When all these velocities are plotted out in a diagram as i 1400 EARLY AND LOWRY THE PROPERTIES OF the figure a curve is obtained which is almost symmetrical 011 either side.According to the well-known principle which is used in determining the ‘( critical volume ” of liquids and vapours the transition-temperature may be determined accurately by ruling a series of horizontal lines across the velocity diagram reading off the temperatures given by the intercepts of these lines with the curve and working out an average value for each pair of tempera-tures. These averages as set out in table V below are as follows: 32.09O 32*11” 32.13O 32*13O 32*12O 32-11° 32.11O. The mean value 32*11° is much more accurate than any of the values deduced either from arrests in the heating and cooling curves of the salt or from merely qualitative observations with the dilatometer and is probably correct to within a few hundredths of a degree.For practical use the second decimal may be omitted and the transition-temperature given to onetenth of a degree as 32.1”. (d) Form of Velocity Cu.rue.*-Whilst the use of the dilatometer in determining transition-temperatures has been a well-known standard method for more than twenty years and has been applied repeatedly when the change of state is too slow to be followed by the thermal method (see especially E. Cohen’sr experiments on the allotropy of metals); the complete velocity curve shown in the figure does not appear to have been plotted in any of these cases. The most striking feature of this curve is its complete symmetry as proved by the constancy of the temperatures shown in the last column of table V which only vary over a range of +0*02O.This complete symmetry which could not be predicted makes it possible t~ locate the transition temperature very accurately and i t is doubtful if any other method is capable of giving equally exact results; for comparison it may be noted that the use ol the “recti-linear diameter ” t o determine critical volumes depends on the existence ol a skew-symmetry oiily in tlie curve of specific volumes for the liquid and saturated vapour. In view of the regularity of the curve it is of interest t o inquire into its mathematical form. The data now recorded can be expresed by the equation f ( t - to) = log,*( .\/ u+ l), where t =temperature, to = transition temperatur,e, v =velocity of change (percentage changed per hour), Added 6/12/19 AMMONIUM NITRATE.PART r. 1401 The agreement of this formula with the experimental results is shown in the following table : 21. 1 2.5 5 10 20 30 40 -J(t - to) obs. 0.33" 0.4 1 0.50 0.62 0.72 0-79 0.86 r t ( t - to). calc . 0-30" 0.41 0.51 0.62 0.74 0.81 0.86 (e) 9no77zctlies i n the Heating mid Coolitig C'urves.-At a very early period in the investigation it was noticed with some surprise that very steady arrests of heating and cooling might occur a t temperatures which were obviously not exact transition points, since they were scattered over a considerable range on either side of a mean value which could be regarded provisionally as the correct transition-temperature.This was an anomaly for which no explanation could be suggested a t the time. General experience in such matters has shown that when deal-ing with rapid transitions such as are observed in iron a t 895O and 766O the arrest in the cooling curve becomes blurred when-ever liberation of latent heat fails to compensate for loss of heat by cooling; a lowered arrest point is therefore nearly always revealed by its sinuous character. I n dealing with ammonium nitrate many sinuous arrests have been observed a t temperatures either below or above the real transition point but in several cases the arrests recorded a t these lower or higher temperatures were perfectly sharp. These observations can all be interpreted in the light of the data now given for the velocity of change of state.Thus in the first place i t is obvious that since this velocity of change is imper-ceptible from 31*B0 to 32*3O there can be no marked liberation or absorption of latent heat and no arrest in the cooling or heating between these limits. Outside these limits of temperature the flow of latent heat may arrest the cooling or heating but the con-ditions are such that the arrest point cannot be regarded as a fixed temperature but must be considered as a variable tempera-ture depending directly on the rate of heating or cooling. Thus, taking the latent heat of the transition as 5.02 calories per gram, and the velocities of change of state as recorded in table IV the temperatures of arrest for the different rates of coding and heating shown in column 2 of table V are given in columns 3 and 4 of that table TABLE IV.Velocity of Change of State above and Motion of meniscus in mm. per 10 minutes. Temperature.* M A 31.2" 31-4 31.5 31.6 31.7 31.8 32.0 32-2 32.3 32.4 32.5 32.6 32.7 32.8 33.0 33.8 8.4 11.0 1.4 0.8 10.5 -6.5 7-0 -13.7 4-8 9.0 t8-4 21.7 25.0 20.0 9.0 8-3 2.1 2-2 1.2 1.9 0.8 0-4 1.1 0.3 0.1 0.1 0-5 0.0 1.2 0.3 1-7 1.6 4.7 3.0 7.6 3.5 7.9 7.0 ts-0 77.7 22.9 23.8 21.0 -8-2 8.1 3-0 3.7 1.9 2.0 0-4 0.7 0.2 0.0 0.0 0.0 0.0 0.0 0.7 0.3 0.1 0.4 2.3 2-2 2.3 2.8 5.7 5-0 t7.2 t7.2 M.3 24.4 -7.5 4.2 2.2 0.7 0.0 0-0 0.0 0-2 0.2 1-7 2-5 4.8 t7-2 24.1 - - -7.5 - -4-5 4.5 4.3 2.5 2-4 2.4 0-6 0.7 0.6 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0-1 0.0 0.3 0.2 0.2 0.2 0.3 -1.5 1.5 1.4 2.0 1-7 1.9 4-0 4.0 4-0 @-8 f6-8 fl-4 - - ---4.7 2.4 0.6 0.0 0.0 0.0 0.2 -1.4 1-7 4-0 fl.3 -* Correction at each temperature = + 0.03" AMMONIUM NITRATE.PABT 1. 1403 TABLE V. Velocity of Change of State of Ammonium Nitrate at Diferent Temperatures. Per cent. changed per hour. 1 2.5 5 10 20 30 40 Rate of cooling or heating. Calories per gram per hour. 0.050 0.126 0.251 0-602 1.00 1.51 2.01 Temperature (arrest-points). - Cooling. Heating. 31.76' 32.42' 31.69 32.62 31-63 32.63 31.61 32.75 3140 32.84 31.32 32-90 31.25 32.96 Average temperature.32.09' 32.1 1 32.13 32-13 32.12 32-1 1 32-11 Mean 32.11" The arrest points shown in the table are the temperatures a t which the latent heat liberated by a change of state proceeding with known velocity would exactly balance the heat gained or lost by radiation conduction etc. Imperfect conduction of heat in the mass may displace these temperatures still further from the true transition point and a t the same time destroy the sharpness of the arrests. The latter phenomenon is observed also in the transitions a t 84O and a t 125O where sinuous arrests a t abnormally low. temperatures are frequently observed but the velocity of change of state a t these two points is so much greater that a pro-longed or " horizontal '' arrest has never been observed except a t a temperature agreeing very closely with one or other of these transition points.Summary. (1) The freezing point of ammonium nitrate is very sensitive to the influence of traces of moisture; by careful purification and drying it has been raised to 169-6"-(2) The highest of the transition points has been determined from the arrests of temperature on heating and cooling the aalt when suspended in a liquid of similar density. The temperatures recorded were on heating 125'24O; on cooling 125-19O; mean value 125.2". (3) The second transition point was determined by the same method the observed temperatures being on heating 84'20O; an cooling 84.20° ; mean value 84.2". (4) The lower transition point cannot be determined in this way as the change of state is too slow to be detected over the range fzom 31'8O to 32'2O. It was determined dilatolmetrically by measuring the rate of change over a range of temperatures; the VOL. axv. 3 1404 KING THE PRODUCTION OF curve of velocities proved to be symmetrical and the transition-temperature was therefore found by taking the average of pairs of temperatures a t which the change proceeded with equal velocities in opposite directions. I n this way the transition-temperature was fixed a t 32.1". (5) The form of the velocity curve for the change of state can be expressed by the simple empirical equation f ( t - to) = k log, ( 4; + 1). (6) The arrest points due to the change of state a t 3 2 ' 1 O never coincide with this temperature the arrest point being determined b y t h e rate of loss or gain of heat. A table is given showing the temperatures of arrest on oooling or heating a t fixed rates expressed in calories per gram per hour. GUY'S HOSPITAL, LONDON S.E. 1. [Recci.ved October 13th. 1919.