]\gt ; Note on the Specific Heat of Water .
By W. R. BOUSFIELD , M.A. , K.C. , F.R.S. ( Received June 16 , 1917 .
) In a paper on the\ldquo ; Specific Heat of Water and the Mechanical Equivalent of the Calorie \ldquo ; *were given the figures from to C. which resulted from our investigation .
These figures agreed very fairly with those of some other observers , but differed considerably from the corl.ected figures of Callendar and Barnes .
In our curve there was a minimum value of the , specific heat at about C. , and a rapid rise afterwards from to of nearly the same magnitude as the fall from to .
In the curve of Callendar and Barnes the minimum was not reached ti1138o C. , and thereafter the rise was much slower .
Callendar soon after deyised a novel method of " " continuous mixture\ldquo ; by means of an ingeniously devised heat exchanger , and applied this apparatus to test his former results in this range between and .
Water at about 10 C. from a heater was passed through the heat against water from a cooler at from to C. The result of these experiments indicated that the ratio of the mean specific heat from to 10 C. to the mean specific heat from to C. reed within 1 in 5000 with the ratio derived from Callendar 's formula based on the figures of Callendar and and Barnes .
These results of this new method were published in Callendar 's Bakerian Lecture in 1912 , S and were made the basis of a criticism of our results and those of other observers .
The value of our results in the which compares with Callendar 's new method depends entirely upon the reliability of the correction for evaporation within this range , which is shown graphically in fig. 8 of our paper .
A glance at that figure shows that it is not till the temperature of is passed that our correction for eYaporation becomes unreliable .
We pointed out , with reference to this correction , that : " " Between and there appears to be considerable uncertainty , especially from to C. This may be partly owing to differences in the amount of air contained in the.water , which tends to come off at about this temperature ; but it is probably due to some variability in the amount of escape of moisture through the holes in the obturator as the vapour pressure rises * W. R Bousfield and W. Eric Bousfield , ' Phil. Trans , vol. 211 , p. 199 ( 1911 ) .
See figs. 10 and 11 , .
cit. , p. 237 .
See Barnes , ' Roy .
Soc. Proc , vol. 82 , p. 390 ( 1909 ) .
.8 'Phil .
Trams , vol. 212 , p. 1 ( 1912 ) .
Mr. W. R. Bousfield .
We have therefore no hesitation in accepting the superior accuracy of the results of Callendar 's new method in this region .
But Callendar 's new figures have only a remote bearing on the values in the lower range and the true position of the minimum .
To test the values in the from to the new method has not been applied , and the difficulties which it was specially designed to meet at higher temperatures hardly exist in the lower range .
Over this range the vapour pressure of water is so small that an accurate correction for evaporation can be easily made .
The figures which we obtained were closely in line with those of other observers , and the question whether they are more or less correct in this range than those of Callendar and Barnes still appears to be an open one .
According to our corrected figures , the mean specific heat from to is almost exactly equal to that from to , which gives the minimum value in the neighbourhood of .
It should not be difficult to apply Callendar 's new method to determine this ratio .
In the meantime , it is desirable to point out what are the outstanding differences in this lower range , and to refer to some new experiments confirmatory of our former figures .
The salient differences in this range of temperature are as follows:\mdash ; 1 .
The fundamental figure which was determined by our experiments was the total heat of distilled water , air freed , and having a conductivity of only , between and .
The value obtained for the mean specific heat of water from to was joules .
The corresponding figure which results from Callendar and Barnes ' observations is 4 .
2 .
The minimum value for the specific heat of water according to our experiments is at , and according to those of Callendar and Barnes at 3 .
The specific heat of water at C. deduced from our experiments was joules .
That deduced from the experiments of Callendar and Barnes was joules .
As to the first point of difference the following observations may be made : The experiments on which our figure was obtained were fairly large scale experiments , there being 3litres of water in the calorimeter .
They were continuous-flow experiments , into which entered no consideration of the capacity of the calorimeter .
They were the result of three sets of experiments , each set of experiments having an ascending order of agreement Note on the Speci .
Heat of between individual experiments .
Before the last set of experiments the electrical standards and the standard thermometers had all been recently and very carefully re-calibrated at the National Physical Laboratory .
The individual results of the last set of experiments were : No. of experiment .
Inflow .
Outflow .
Value of Mean 41785 \mdash ; The mean of these was the value for tap water .
The difference between individual values and the mean was less than 1 in 4000 .
The range of temperature being to , it was estimated that there might be an error of in the temperature readings of any of these intervals .
The result shows that the mean error was less .
There was no chance of any zero error creeping into the results .
These experiments , each of which involved the use of a large of water , were made on " " tap water The resulting figure was used to determine the capacity of the calorimeter .
A series of fifteen subsequent experiments on air-freed distilled water over the same range showed that its mean specific heat was slightly higher , viz. , from to .
Subse- quently the total heat of water for each of the three intervals , , was determined by four series of experiments , and the sum of the total heats so determined was precisely the same as that directly determined , again giving for the mean specific heat from to It is suggested that there may be some hidden systematic error in the method .
This is , of course , possible , but it seems to be equally possible in the continuous-flow method with platinum thermometry on which the figures of Callendar and Barnes were founded .
As a practical unit for calorimetric work the mean specific heat from to still stands preeminent , since it can be readily reproduced at all seasons of the year , and easily applied to ascertain accurately the capacity of any calorimeter , figure which is necessary for nearly all calorimetric work on solutions , where the quantity of solution required forbids the continuous-flow method in most cases .
As to the second point of difference , the position of the minimum , the VOL. XCIII.\mdash ; A. 3 Mr. W. R. Bousfield .
following observations may be made : The course of our curve from to with its minimum at about ) is in general reement with the cur.ves of Ludin , Bowland and Bartoli and Stracciati , whilst the course of the Callendar and Barnes ' curve with its minimum at about differs from all .
The results of a paper on the specific heat of certain solutions presented simultaneously with this note confirm our position of the minimum very substantially.* The old Dewar vessel having been broken it was required to determine the mean capacity of a new and heavier vessel over the intervals .
A completely different method of regulating the electric current was adopted .
In the former experiments the current was kept constant by the use of an ampere balance and simultaneously of a battery of standard cells .
In the present experiments the current was so as to keep the energy supply constant by means of a Kelvin watt balance .
Moreover , the system of corrections for small heat gains or losses was entirely changed .
The capacity was determined by a series of twelve experiments which are set out in the simultaneous communication .
For any interval to , the total heating of the calorimeter and its contents is given by where is the weight of water , the mean specific heat of water for the interval , the mean capacity of the calorimeter for the interval , the electrical watts , the mean rate of rise of temperature per minute and the correction for small heat gains or losses .
From these experiments the mean values for the quantity on the right-hand side were determined as Joules . . .
13382 13401 .
The mean value of was The mean values of derived from the former paper , after making a minute correction for the slight variation of capacity of the calorimeter which arises from the expansion of the water from to were 4 .
4 .
4.175 whilst the responding values of from Callendar and Barnes ' figures are .
, This paper will appear in the ' Philosophical Transactions .
' Note on the Specific Heat of Water .
The respective capacities derived from the first set of figures are whilst those derived from the second set of figures .
As the specific heat of glass steadily increases with rise of temperature it is obvious that the second set of capacities is inadmissible .
The ularity of the second set clearly arises from the shifting of the minimum which these figures involve and the regularity of the series of capacities derived from our figures supports the view that the minimum is in the neighbourhood of which ives almost equal values for and and not diminishing values .
A further confirmation results from a consideration of the actual values of the specific heats of the KC1 and solutions .
The phenomenon of the minimum value of the specific heat of water undoubtedly arises from the fact that the large proportion of ice-molecules contained in water at the lower temperatures is considerabl .
reduced with rise of temperature , whilst a little later the increasing proportion of steam molecules tends again to increase the specific heat .
In saturated solutions of KC1 and the proportion of ice and steam molecules is yreatly reduced , and the solvent becomes mainly simple dihydrol .
The ures for these solutions indicate a nearly straight line law from to between specific heat and temperature .
The fact that the temperature-specific heat curves for saturated solutions of KC1 and come out as nearly straight lines can hardly be a mere coincidence and it entirely depends on the right relative values for the capacities for the three intervals .
Until the course of the specific heat curve from to can be considered as definitely settled there must remain a doubt as to the value of the calorie amounting to 1 part per 1,000 .
The reasons for as the standard calorie the mean specific heat over an interval which undoubtedly includes the minimum point upon either view and which is easily reproducible were stated in the former communication .
* The interval from to satisfies all the practical conditions and there remains a difference of over 1 part per 1000 between our value for the interval and that which results from Callendar 's latest figures .
It may be hoped that Prof. Callendar may be able to apply his new method to a direct test of the values over this range . .
cit. , p. 42 .

