]\gt ; [ 123 ] III .
On Evaporation \mdash ; Part II .
A Study the Thermal Properties of Alcohol .
WILLIAM RAMSAY , .D .
, and SYDNEY YOUNG , D.Sc .
Communicated by Professor G. G. STOKES , Sec. Received April 18 , \mdash ; Read May 7 , 1885 .
[ PLATES 3-7 .
] 1 .
THE density of gases is found to increase as the temperature falls towards their condensing point .
This may be explained by one of two theories:\mdash ; ( 1 ) That complex gaseous molecules are formed in sing numbers as the temperature falls ; these complex molecules consisting of congeries of the simpler molecules known to exist in gases ( see PLAYFAIR and WANKLYN , Trans. Roy .
Soc. Edin .
, xxii .
, ( 3 ) , p. 441 , and ' Annalen , ' 122 , p. 245 ; also NAUMANN , 'Annalen , ' 155 , p. 325 , and ' Thermochemie , ' pp. 86 et seq ( 2 ) That mutual attraction of the molecules , caused by their greater proximity , comes into play , and reduces their mean distance from each other .
The second of these hypotheses involves a deviation from AVOGADRO'S law ; while the first does not .
Similarly , with those substances in which dissociation takes ) lace in the gaseous state , the density of the mixture of gases increases with Iall of temperature , and this is doubtless caused by the increase in number of compound molecules .
That this is the case with nitrogen peroxide can hardly be questioned , from the different physical properties of the gas under different conditions of temperature ; and the phenomenon is also well seen in the case of tertiary amyl bromide .
It may be pointed out that both these causes may operate at the same time .
It has occurred to us that light may be thrown on this question by a study of the properties of three representative compounds:\mdash ; ( 1 ) What may be called a normal substance ; i.e. , a substance in which no chemical dissociation is generally supposed to occur .
Ethyl alcohol has been chosen as a representative of such a type .
( 2 ) A substance the vapour-density of which shows marked abnormality , and which is suspected by many chemists of forming complex molecular groups .
It is generally supposed that acetic acid , , tends to form double molecules of the formula and thab its abnormal vapour-density is to be explained on this hypothesis .
Hence we have chosen it as a subject of experiment .
( 3 ) It is hoped that it may be possible to effect a similar of a sub124 ce which undou cecPROFESSOR WRAMSAY AYOUNG the dissociation of stance which undoubtedly dissociates into dissimilar molecules , or the dissociation of which is undoubted from some change in its physical properties .
2 .
Such a study as we propose involves the determination , within the largest : possible range of temperature , of the expansion of the liquid and of its compressibility ; of the density of the vapour when unsaturated as well as when in contact with its liquid ; and of the heat absorbed in producing change of state from liquid to gas .
We have already accumulated a number of data dealing with a portion of this problem ; and as they are necessarily very voluminous , we deem it advisable to publish those bearing on the thermal constants of alcohoL 8 .
It is proposed to reserve until later a detailed description of the apparatus with which this research was executed .
It will be sufficient here to state that it is a modification of ANDREWS ' apparatus , provided with two air-gauges , one for measuring low , and the other for measuring high pressures .
The temperature of each gauge was accurately known , for they werejacketed with running water in which a thermometer was immersed .
These gauges were graduated in millimetres and calibrated by weighing with mercury , and were filled with air dried with phosphorus pentoxide .
In calcuJating pressure , the following data were required:\mdash ; ( 1 ) The temperature and pressure of the air when the gauge was filled .
( 2 ) The temperature of the air when readings were taken .
( 3 ) The coefficient of compressibility of dry air , taking into account the deviation from law , as determined by AMAGAT , ' Comptes Rendus , ' 99 , p. 1153 .
( 4 ) The error of the gauges , corrected by calibration .
( 5 ) The difference in height of the columns of mercury in the gauges and in the tube containing the liquid .
( 6 ) The vapour-pressure of mercury in the experimental tube .
At the highest temperature employed , , it amounts to 68 mms .
, or about of the total pressure measured .
For this , in many cases , we have applied no correction , for it is doubtful whether the mercury can give off vapour when its surface is covered by a long column of liquid .
From experiments on the vapour-pressure of bromo-naphthalene at at which the vapour-pressure of mercury is 55 mms .
, it was noticed that a considerable time was required for the vapour to rise through a comparatively short column of liquid bromo-naphthalene .
It is , therefore , probable that the full pressure of mercury vapour was not exercised : and the possible error is reduced below the experimental error of reading at such high pressures ; but in the case of unsaturated vapour-densities , this correction was introduced .
4 .
The experimental tube , containing the alcohol , confined by mercury , was 200 .
long , and was graduated in millimetres , and calibrated by weighing with mercury .
It was made of lead-glass barometer-tube , with an internal diameter of about 2 mms .
The coefficent of expansion of this ] was determined between and , and was found to be per degree .
In correcting for the meniscus of the mercury , the meniscus was regarded as a hemisphere .
ON THE THERNIAL PROPERTIES OF ALCOHOL .
A similar correction was made for the surface of the alcohol , no apparent change taking place until its temperature was raised to within a few degrees of its critical point .
At those temperatules it was found that the error was inappreciable .
The experimental tube was heated by the vapours of pure liquids , boiling under pressures which could be varied at will .
The vapour-pressures of these liquids have been accurate]y determined , temperature bein measured with a mercury thermo- meter .
Several measurements were carried out with an air-thermometer , and the necessary corrections were applied to reduce the readings of the mercurial thermometers used to those of an air-thermometer .
The details of this investigation have been given in the Trans. Chem. Soc. , 1886 , p. 37 ; the results which are there published constitute a guarantee of the accuracy of the determination of temperature .
We can state with confidence that at the highest temperature employed the absolute error does not amount to two or three-tenths of a degree ; and the temperature can be altered with certainty through by alteration of the pressure under which the liquid is boiling .
This is , we believe , the only method by which a long tube can be uniformly heated ; and by which the temperature can be easily and quickly altered .
All the temperatures given are , therefore , those of an air-thermometer .
5 .
It is unnecessary to give details as regards reading , calibration , \amp ; c. , for all our results .
But as it may help to explain our method of procedure , one instance shall be given , and the method may be held to apply to all the other results .
The temperature on an air-thermometer at which aniline boils at an observed pressure of , the temperature of the gauge being ( equivalent to mms .
at , is .
At this temperature , the following readings were made : Reading in experimental tube of upper surface of mercury , 210 .
Reading of upper surface of liquid , volumes corrected for meniscus : volume of liquid plus gas .
; volume of gas .
Add correction for expansion of glass ; for higher volume , ; for lower volume , .
Corrected volumes ; ( 2 ) .
Volume of liquid ; volume of gas .
The volume was subsequently reduced , until ] a minute bubble of vapour was left uncondensed .
It was thus made certain that the liquid , as such , was not being compressed .
The corrected volume was then found to be The weight of the liquid in the tube was .
; this latter number was found from several readings of the volume at known low temperatures , and the known specific gravity of alcohol at those temperatures .
The specific gravity of alcohol under pressure corresponding to that of its own vapour at .
From this number is deduced the weight of the liquid .
The weight of the vapourtotal weight\mdash ; weight of liquid The weight of 1 cc .
of vapour therefore equals its weight divided by its volurne Vapour density :Weight of 1 at and 15,163 mms .
pressure Hence Pressure :Reading on high-pressure gauge , .
Temperature of gauge , Ratio of total volume of gauge tube to that of compressed air , the correction for the meniscus of the mercury in the gauge being applied , : 1 .
Ratio corrected for temperature and pressure and for temperature at which gauge was filled atmospheres mms .
Add difference in heights of mercury in gauge and in experimental tube mms .
From this example an idea may be gathered as to the methods of reduction employed , and of the corrections introduced .
6 .
We now proceed to give results corrected by the method above explained .
PRESSURES of Vapour .
7 .
As it is impossible to distinguish between liquid and gas at the critical point , so it is impossible to give the vapour-pressure at the critical point .
The pressures * Correction for divergence of dry air from law , ON THE THERMAL PROPERTIES OF ALCOHOL .
recorded will be given when treating of the compressibility of the liquid .
They remained constant within the limits of reading during a considerable reduction of volume .
8 .
The vapour-pressure of alcohol has been measured by REGNAULT up to On examining his tabulated results by the method of differences , it was found that the third differences show ularity at and above ; on taking the fourth differences minus quantities begin to appear .
We find that our numbers to some extent confirm this peculiarity of REGNAULT'S results .
In order to calculate the pressures corresponding to even temperatures , the difference for , i. e. , about below and about above the temperature requiring reduction was taken ; and by dividing by the interval of temperature , the mean change for].o was found .
This was applied to reduce the pressure at the observed temperature to what it would be were the numbers representing temperature multiples of The error introduced by this somewhat crude method must necessarily be small , inasmuch as the required change of temperature seldom exceeds 1o From measurements , described in a later part of this paper , the vapour-pressure at was found to be mms .
From this number , and from those at , and , constants for a formula of the type suggested by BIOT were calculated .
The formula .
The constants ; is negative ) .
*'Memoires de l'Institut , ' vol. xxvi .
, p. 374 .
'Comptes Rendus , ' vol. 1 .
, 1860 , p. 1067 .
9 .
Specific Specific Volumes .
In order to ascertain the specific gravity of alcohol , it was necessary to know the volumes at various temperatures , and the total weight of substance .
As the weight was not the same throughout , losses having occurred through breakage , it was not possible to obtain the weight directly .
It was , therefore , necessary to deduce the weight from the volume and the known specific gravity at low temperatures .
The results given by different observers are , however , not uniform We quote some of these ; in all cases comparison is made with water at as unity .
PIERRE ( ' Annalen , ' 56 , p.142 , and ' Annals , ' 15 , p. 325 ) KOPP ( ' Annalen , ' 64 , p. 213 ) MENDELEJEFF ( ROSCOE and MENDELEJEFF ( ' Annalen , ' 119 , p. 7 ) PERKIN ( Chem. Soc. Journ. , 45 , p. 465 ) MENDELEJEFF ROSCOE SQUiBB ( Chem. News , 51 , p. 21 ) .
In calculating our results , the number was taken ; it will be seen that the mean of the last three results , which are probably the best , is .
The alcohol employed had been frequently distilled over lime , and was finally dried by distillation with a little sodium .
It boiled with perfect constancy .
10 .
During these experiments , an accident occurred , by which we lost the first portion of alcohol ; the tube was therefore refilled .
We shall name the first portion , and the second , in giving data .
11 .
In the following table , the pressure under which the liquid was measured , was in general the vapour-pressure corresponding to the temperature .
The asterisked figures are corrected to this standard from observation on the compressibility .
ON THE THERMAL PROPERTIES OF ALCOHOL .
12 .
Compressibility .
The compressibility was measured by increasing the pressure , and after some time reading the volume of the liquid .
It was noticed that heat was developed by compression , hence time allowed for the heat to dissipate before reading .
The following are the results : of 1 gram .
ON THE THERMAL PROPERTIES OF ALCOHOL .
The graphic representation of these numbers shows that the compressibility at low temperatures is nearly constant through a large range of pressure ; but at high temperatures it varies , decreasing rapidly as pressure rises , and near the critical point , as observed by ANDREWS , the liquid is much more compressible than a gas .
13 .
By plotting these results in curves , we are enabled to give the volumes occupied by liquid alcohol at constant pressures , with variation of temperature .
To do so , however , required a somewhat tedious process .
The actual results were plotted on curve paper , and by means of engineers ' curves ( circles of different radii ) , lines were drawn , occupying as nearly as could be judged a mean position among the observed points .
From these curves the volumes for certain definite pressures were read off , intervals of 5000 mms .
at low temperatures , and 2500 or 1250 mms .
at temperatures being chosen .
From these numbers fresh curves were constructed , showin relations between temperature and volume for a series of pressures .
These curves were also smoothed , and the results obtained by their means retransferred to the first set of curves .
The curves published in Plate 8 show these doubly-smoothed curves , and the actual observations are also shown by circles , which will enable the amount of error to be estimated .
The temperatures of the isothermals are written ON THE THERMAL PROPERTIES OF ALCOHOL .
on the horizontal lines which represent change of volume without increase of pressure , where vapour in contact with liquid is being condensed .
14 .
The temperatures given are those read directly , with two exceptions , viz. , was altered to , and was altered to .
The reason for this alteration was that in drawing the isobaric curves , deduced from the isothermals , the observed points at those temperatures differed from the curve in every case by the same amount .
This error is really a very small one at such high temperatures , and may depend on some mistake in reading the pressure under which the bromonaphthalene used as a jacket was boiling .
It is to be remarked that by this double method of smoothing any great irregularity of temperature would at once be detected ; this is shown by the corrections above referred to .
ON THE THERMAL PROPERTIES OF ALCOHOL .
16 .
A set of determinations have been made by ZUK Rusk .
Chim .
( 2 ) , p. 239 ) of the volumes of alcohol under the critical pressure observed by SAJOTCEWSKY of atmospheres mms We are now in a position to compare our results with his .
ZUK adopts the specific gravity at , a number close to that adopted by us , .
The comparison is as follows : \mdash ; It is evident that there are considerable differences between the two sets of results .
These are probably to be accounted for by differences in temperature .
It should be remarked that ZUK employs as the tempelature of the critical point The formula used by ZUK is one given by AVENARIUS , .
:\mdash ; .
As this formula is entirely empirical , there is nothing to be gained by its use , although it appears to give an approximation to the found results .
17 .
We are now able to giye the specific volumes and specific gravities of liquid alcohol at definite intervals of temperature and at pressures equal to the vapourpressures at those temperatures .
They are given in the following table .
They are read from a smoothed curve epresenting individual observations of temperature and volume , and checked by means of the curves representing compressibility for known rise of pressure at constant temperature .
In Plate 3 , the observations shown by circles determine the direction of each curve .
The curve connecting their terminals , which exhibits relations between the volume of the liquid , the temperatures , and the pressures equal to the vapour-pressures at those temperatures , was constructed by COlTecting the volumes when not read at the true vapour-pressures to the true volumes at those pressures .
The curve was drawn through the points thus obtained , and it was seen that the points of intersection of these curves confirmed the position of points on a curve representing the relations between volumes and temperatures the vapour-pressures corresponding to the temperatures .
We would suggest the name " " orthobaric\ldquo ; as applicable to such a curve , and ' ' orthothermic\ldquo ; to denote the curve representing relations between volumes and pressures at temperatures responding to the vapour-pressures of the substance .
18 .
It would be possible to obtain the volumes of 1 gram of the liquid at definite temperatures and at corresponding vapour-pressures directly from the " " orthobaric\ldquo ; curve , or by constructing new isothermals , by means of the isobaric curves , for such temperatures and reading the points of intersection of such curves with the " " orthothermic\ldquo ; curve .
Since the results for the temperatures of observation were practically identical , it was judged sufficient to read them directly from the " " orthobaric\ldquo ; curve .
19 .
The numbers thus read are as follows:\mdash ; SPECIFIC Volume and Specific Gravities of Alcohol at Definite Temperatures , and at the Vapour-pressures corresponding to these'Temperatures .
20 .
For the sake of comparison , we give HIRN'S ( ' Annals , ' ( 4 ) , 10 , p. 32 ) and MENDELEJEFF'S ( loc. cit. ) numbers .
MENDELEJEFF . .
Found ( R. and Y. ) HIRN . . . . .
Found ( R. and Y. ) ZUK also , .
cit. , gives numbers for the volumes of alcohol .
These have already been considered .
Specific Volumes and Specific Gravities of Saturated Vapour of Alcohol .
21 .
It has been explained in S5 how the volume of the saturated vapour was measured , and the various corrections introduced .
We now proceed to give the experimental numbers .
* As some confusion has been introduced by the use of the phrase ' specific volume\ldquo ; as synonymous with ' ' atomic\ldquo ; or " " molecular volume it should here be stated that the expression is employed to denote the volume of 1 .
of the substance .
ON THE THERMAL PROPERTIES OF ALCOHOL .
In calculating the weight of 1 cub. centim. of saturated vapour , it has been pointed out , in S5 , that this constant depends on the specific gravity of the liquid .
Now the specific gravity of the liquid may be taken , ( 1 ) as directly read , or ( 2 ) from the smoothed orthobaric curve .
Both of these have been made use of , and the results are compared in the following table:\mdash ; The chief differences in the numbers in these columns are at low temperatures , where a smaJl alteration in the specific gravity of the liquid causes a large change in the weight of 1 .
of saturated vapour .
22 .
From the data already given , the densities compared with hydrogen under similar conditions of temperature and pressure were calculated .
In these calculations the data given in columns 1 , 2 , and 3 of the preceding table have been employed , and the corresponding vapour-densities are similarly numbered .
The vapour-densities found were then plotted on a curve , and the 4th column shows the smoothed results .
DENSITIES of Saturated Vapour 23 .
It is evident , on inspection of this and the preceding table , that the results at lower temperatures are not nearly so accurate as those at higher .
The cause of this is that an error in the specific gravity of the liquid , which would have almost no on the vapour-densities at higher temperatures exercises a serious influence , for it will be remembered that the weight of the vapour is deduced from the total weight of substance and the weight of the liquid , while the weight of the liquid is deduced from its specific gravity and its volume .
As the difference between the total weight and the weight of the liquid is very small , the ratio of the error in the weight of the liquid , and therefore of the vapour , to the total weight of vapour is a large one .
24 .
The preceding table also shows that the vapour-density decreases very rapidly at first , and afterwards more and more slowly with fall of temperature , from the highest temperature to about .
Below that temperature the results were not certain enough to enable us to decide positively whether the density of the saturated vapour remains constant during a considerable fall of temperature or whether it begins to increase slowly .
Determinations shall be described further on which elucidate this point .
It is one of extreme importance , for an increase of vapour-density ON THE THERMAL PROPERTIES OF ALCOHOL , would imply the probable formation of complex molecular groups ; that this is the case with acetic acid vapour will be shown in a subsequent memoir .
25 .
In concluding this part of the subject we give a table taken from the curves , showing specific volumes , specific gravities , and vapour-densities at definite intervals of temperature .
SPECIFIC Gravity , Specific Volume , and Density of Saturated Vapour at definite Intervals of Temperature .
dens .
The bracketed numbers express the limits between which these constants appeared to fall .
26 .
of of Alcohol .
From the foregoing data it is possible to calculate the heats of vaporisation of alcohol by mealR of the formula Heat of vaporisation .
and Volumes of saturated vapour and liquid respectively .
Absolute temperature .
Mechanical equivalent of heat .
Rate of change of pressure , measured in grammes per square centimetre .
27 .
This rate of change was ascel.tained from the vapour-pressures of alcohol , calculated by BIOT'S formula , two points being chosen , above and below the temperature required , and the difference between these pressures was multiplied by 5 .
From to 23 inclusive the values were read from a curve constructed to show relations of to temperature ; the number at is calculated , but at and the curve was again employed .
The results are given in the following table:\mdash ; of for definite Temperatures .
28 .
VALUES of L. refer to S 40 .
) was taken as 42500 .
square centimetre .
ON THE THERMAL PROPERTIES OF ALCOHOL .
29 .
It will be seen from these numbers that the values of increase regularly to with fall of temperature .
They then appear to remain approximately constant .
It is at this temperature also that the vapour-dellsities become uncertain , and show a tendency to rise .
From the nature of the equation used to obtain the values of any discrepancy in the vapour-densities will affect the values of L. At temperatures below l80o , as explained in S 23 , the error introduced becomes considerable , and the experimental results are evidently no longer trustworthy .
But above this temperature the ratios of the specific gravity of the liquid to that of the gas becomes sufficiently small to give accurate results ; and the accuracy is confirmed by the regularity of the deduced curve .
30 .
There are already data as regards the heat of vaporisation of alcohol .
ANDRE ws ( Chem. Soc. Journ. , I. , p. 27 ) gives the value of for , under a pressure of 760 as cal .
FAVRE SILBERMANN ( ' Annales ' ( 3 ) , 37 , p. 465 ) give at the boilingpoint under normal pressure .
REGNAULT ( ' Memoires de l'Institut , ' xxvi .
, pp. 819 and 262 ) gives a number of values of the total heats , from which the heats of vaporisation may be calcuiated by subtracting the specific heats of the liquid , calculated according to the formula given by Sp. Ht .
These numbers are as follows , and we append the volumes of saturated vapour and the resulting vapour-densities calculated from his numbers .
They are shown graphically on the accompanying curve in Plate 4 ( 1 ) Total h Latent h As alcohol cannot without dissociating possess a vapour-density less than 23 , the numbers below are manifestly wrong .
ANDREWS ' measurement gives a vapour-density of ; and FAVRE and SILBER142 sPROFESSOR WRAMSAY A es calculations our ; both of which agree well with our results .
In these calculations our values of were employed .
31 .
HERWIG gives a ] by which the volume of 1 gram .
of saturated vapour of various liquids may be calculated from the known value of PV when constant .
This formula is Where is the volume of 1 gram ; the mass of substance ; , 273 ; , temperature centigrade ; and , vapour-pressure at temperature .
Although this formula does not hold in the least at high temperatures , we give the results for alcohol calculated by HERWIG , at temperatures within the limits of which HERWIG states that his law holds good .
( HERWIG states that the alcohol he used contained a trace of water .
) If the vapour-densities were continued to lower temperatures they would manifestly fall below the lowest possible .
Had our values of been substituted for those of REGNAULT , the discrepancy would have been even greater .
The numbers of ANDREWS , and of FAVRE and SILBERMANN , are fairly concordant , and appear to indicate a vapour-density of to at normal pressure ; and it will be seen that a continuation of the curve obtained by us might hit this number at It appeared desirable to confirm this view by further nents in which different methods were employed .
32 .
Densities of Unsaturated Alcohol Vapour .
For higher temperatures the process was essentialy the same as that already described .
A much smaller amount of alcohol was used , so that it was possible to convert it into gas , and to increase volume until its density became normal at a high ON THE THERMAL PROPliRTIES OF ALCOHOL .
temperature .
Three sets of experiments were made .
In the first too large an amount of alcohol was taken to enable the vapour to acquire normal density , even with the largest volume which could be given .
The volume of the liquid was therefore measured at 22 and , and its weight was calculated from the data obtained , and from the determinations of the specific gravity of alcohol which have already been described .
In the second and third series the density was normal under the greatest volumes given , for the product of pressure info volume remained constant during change of volume .
SERIES I. Weight of Substance .
I. Volume at . . . .
Sp. Gr .
Weight II .
Volume at Sp. Gr .
Weight .
0 .
0.4958 .
gram .
0.04514 cc .
0.4775 .
gram .
Mean , 33 .
A quantity of the alcohol was removed and a fresh set of determinations made et .
at .
A curve was then constructed from the results of Series I. , showing relation between volume and pressure ; the volumes on this curve corresponding to the pressure on the second series were then read , and the ratios between the volumes in the two series at the same pressures were determined .
By this method the weight of the alcohol used in Series II .
was found to be .
It will be seen that with this smaller quantity of substance the vapour-density at the higher temperatures and largest volumes becomes constant at 23 , and this affords sufficient confirmation of the correctness of the weight given .
ON THE THERMAL PROPERTIES OF ALCOHOL .
NDCCCLXXXVI .
ON THIB THERMAL PROPERTIES OF ALCOHOL .
P. 34,792 3,576 8,722 3,898 4,489 5,308 6,495 7,639 9,257 10,318 11,635 13,367 15,733 17,202 19,082 21 , 190 23,827 27,144 31,636 33,751 35,650 37,556 38,093 3,604 3,759 3,935 4,772 6,089 7,690 10,877 1.3,480 15,759 19 , 108 27,448 32,091 35,055 37,311 38,071 38,961 39,650 3,673 3,999 4,862 6,210 8,628 11,969 16 , 135 19,458 24,504 27,740 32,721 36,030 38,092 39,104 40,048 40,711 40,894 3,690 Vapour density ) 139.7 170.7 6.404 ON THE THERlfAL PROPERTIES OF ALCOHOL .
SERIES IIL 34 .
The amount of alcohol used for the determinations of which the results are given above was too large to admit of values at lower temperatures ; a fresh portion was removed , and the weight of the remainder was ascertained by determining the value of P. V. at a temperature at which it became constant at the highest volumes .
temperature was , and the mean value obtained for P. V. was .
From this the weight of alcohol was calculated to be The following measurements were made : ON THE THERMAL PROPERTIES OF ALCOHOL .
35 .
The results of all these determinations of vapour-density are represented , so far as it is possible without introducing confusion , on Plate 4 ; the ordinates represent pressures , and the abscissae vapour-densities compared with hydrogen under similar conditions of temperature and pressure ; the smaller curve represents the results of the experiments at lower temperatures on a larger scale .
The lines are isothermals ; and the curve connecting their terminal points gives the densities of the saturated vapour at the corresponding temperatures and pressures .
From inspection of this curve it is evident that these results for densities of saturated vapour confirm those obtained by direct measurement at temperatures above ls ; but the previous results are incorrect , below that temperature , for reasons already given .
far as can be judged from these results , bhe density of the saturated vapour of alcohol decreases with fall of temperature , and corresponding fall of pressure .
36 .
But from experiments with acetic acid , this is found not to be the case with that substance .
On the colltrary , the density of its saturated vapour increases rapidly with fall of temperature and pressure .
Hence , it appeared necessary to carry out experiments on the density of saturated alcohol-vapour at still lower temperatures , for it was deemed possible that after reaching a minimum density , fall of temperature would produce an increase .
The apparatus , which will be fully described in an account of the experiments with ether , consisted of vapour-density apparatus , so modified as to admit of alteration of pressure .
The desired temperature was produced by jacketing with running water .
The following results were obtained at :\mdash ; At a volume of 59 cub. centims. liquid was present .
37 .
This method involves three readings of heights of mercury , and calculation of a fourth ; and as an error of .
makes an alteration in the value of P. V. of 6 to 10 units , the results , although irregular , are well within the limits of experimenta ] error .
They show that the product of pressure into volume remained practically constant up to a very small distance from the volume at which condensation began to occur .
38 .
From measurements at the boiling-point of alcohol under atmospheric pressure , an attempt was made to ascertain the weight of the extremely minute amount of alcohol present .
From the mean of these results , 1976 , the weight of alcohol was calculated to be Using this weight to calculate the vapour-density at , the number was obtained .
This result is not unsatisfactory , considering the experimental difficulties ; and , in our opinion , it proves that the density of the saturated vapour is practically normal at low temperatures .
39 .
The densities of the saturated vapour can be deduced from the curves showing the densities of unsaturated vapour ; they are represented by the points of intersection of these curves with the horizontal lines which exhibit change of volume of gas in presence of liquid , while pressure remains constant .
They were read from the smoothed curve given in Plate 4 , and they are as follows:\mdash ; On comparin the results of this table with that previously given in S22 , it will ON THE THERMAL PROPERTIES OF ALCOHOL .
be seen that above ls the agreement is close , but that , as was expected , the results diverge widely at lower temperatures ; above the agreement was perfect .
The values at , and between and are obtained by interpolation on curve 2 , Plate 5 .
40 .
Heats of of Alcohol .
From these numbers the heats of vaporisat , ion were calculated by the method already mentioned .
They are given in the table which follows:\mdash ; ( For temperatures above 210 reference must be made to S 28 .
) 41 .
The relations between temperature and heats of vaporisation are graphically represented in the two curves on Plate 5 , fig. 1 , curve No. 2 , represents this relation for vapour-density .
The points experimentally determined are indicated by circles .
Between and the dotted line shows the probable course of the curve .
It is obvious that the error must be extremely small , and it is probable that the vapour-density attains its minimum at a temperature somewhat below 42 .
The curve representing latent heats is similarly indicated by circles and dots .
That part of it deduced from the interpolated portion of our observationls of yapourdensity seems to point to a constant value for heat of vaporisation below about the temperature at which vapour-density becomes normal being about This would imply that below that temperature the difference between the specific heats of the saturated liquid and saturated yapour is inversely proportional to the absolute temperature ; for in the equation the expres ion vanishes .
* In grams .
per square centimetre . .
As it was difficult to understand the reason of the great divergence between the heats of vaporisation found by REGNAULT and those calculated from our results we it possible that some error in REGNAULT'S determinations of the vapourpressures of alcohol , which were at first used as the basis of our calculations , might be the cause of this discrepancy .
It has already been ticed that the vapourdensities calculated from REGNAULT'S results with heats of vaporisation are impossible ; and by taking the vapour-density as normal the calculated curve of heats of vaporisation approaches as near REGNAULT'S as is possible .
Now , as the vapoursure of alcohol is a datum for the calculation of the value of , and as this value , also that of , occurs in the equation an error in REGNAULT'S determinations of vapour-pressure , which we have at first employed , would have influenced all the foregoing results , and most markedly those at lower temperatures .
We , therefore , thought it able to make a fresh series of determinations of the yapour-pressure of alcohol , and used for this purpose the described in the Trans. Chem. Soc. , Jan. , 1885 .
The temperatures are corrected for the change in volume of the bulb of the thermometer by diminution of pressure , which , with the particular thermometer employed , one graduated in of degrees , by and AMBRA , amounted to an apparent fall of on lowering pressure from 763 rnms .
to 20 mms .
44 .
The results are as follows:\mdash ; 45 .
From these results a curve was constructed , and from it values at definite temperatures were read off They are compared with those given by REGNAULT in the ' Me'moires de l'Institut , ' vol. xxvi p. .374 .
ON THE THERMAL PROPERTIES OF ALCOHOL . .
( R. and Y. ) 46 .
It will be seen from these results that our pressures are uniformly lower than those given by REGNAULT , except at , where there is manifestly a misprint in his figures .
The difference in pressure is approximately constant , and is equal to mm. The value of is considerably changed by this alteration , and the numbers given , except where the contrary is stated , have been derived from the formula , the constants of which were calculated by us for vapour-pressures .
The Critical Point of Alcohol .
47 .
In the ' Philosophical Transactions ' for 1869 , p. 583 , ANDREWS gives a diagram showing , to some extent , relations between the pressure and volume of carbon dioxide at temperatures near its critical point .
From our observations on alcohol we are enabled to sinlilar diagrams on Plate 6 , showing the volumes of 1 gramme .
The first of these ( fig. 1 ) comprises observations between and .
The scale is necessarily a small one .
The isothermal for has beeu mapped , and it is to be noticed that all other isothermals between these limits of temperature would ] between that line and the one representing volumes of saturated vapour .
The umbers deduced from observations are represented by circles .
Fig. 2 shows similar relations at temperatures between 22 and on an enlarged scale , and it also exhibits the compressibility of the liquid .
The hest temperature at which it possible to measure both liquid and gas was above that temperature measurement of gas was uncertain , owing to the slow l.ate at which condensation or evaporation proceeded .
A successful measurement of the volume of the liquid was made at , but it was still possible to distinguish liquid from gas ; at the meniscus appeared to broaden and became indistinct .
It disappeared absolutely on the least decrease of volume , owing to heat developed by compression , and it rea , ppeared on increasing volume .
The apparent critical point was therefore at this temperature , but the pressure remained perfectly constant during considerable change of volume .
A similar phenomenon has been noticed by ANSDELL , and we agree with him that the real critical point must lie somewhat higher .
In the Plate 6 a dotted line has been drawn connecting the curves showing the volumes of saturatedvapourandofliquid .
Thesummitofthissurface , which rPROFESSOR WRAMSAY AYOUNG OLOOHOL nts volumes of gas contact with liquid , correctly drawn , would give the relation between temperature , pressure , and volume at the critical point .
It appears to us impossible to obtain direct measurements of this portion of the curve .
There must , therefore , necessarily be some uncertainty as regards all three .
On Plate 7 the relations between the weight of one cubic centimetre of saturated vapour and liquid are shown .
Curve No. 1 exhibits this relation between and ; and curve No. 2 between and on an enlarged scale .
It is evident , however , from these results that the critical temperature must lie within of a degree above or below ) , that the critical pressure must be nearly 47,700 mms .
, and that the critical volume cannot be far from , one gramme of alcohol at the critical point must occupy about .
; otherwise expressed , the specific gravity at the critical point is nearly NOTE .
After this memoir had been read ( May , 1885 ) , an account of experiments by AMAGAT ( Count .
Rend .
, xcix .
, p. 1153 ) on the compressibility of air was noticed .
reputation and the method he adopted are a sufficient guarantee for their accuracy , and , although these results differ considerably from REGNAULT'S , which , moreover , had reference only to pressures below 20,000 mms .
and were therefore not available for our purpose , we have felt justified in making use of them .
This has necessitated a recalculation of all the pressures , values of , and vapour--densities given in this paper , and has led us to modify some of the conclusions in the memoir as read in 1885 .
It was also thought advisable to use BIOT'S formula , aild to substitute numbers for pressure calculated by its help , instead of by the method of differences was at first employed .
The use of this formula enabled us to obtain more accurate values for than was possible by any graphical means , and , as a result of these alterations , th curve representing heats of vaporisation is much more regular than we at first supposed ; besides , it passes between the points obtained from direct measurement by ANDREWS , and by FAVRE and SILBERMANN .
The diagrams have also been reconstructed so as to represeut accurate]y the corrected numbers .
/ ' / ta 7 \mdash ; - \mdash ; \mdash ; c ) .

