]\gt ; Hon. C. A. Parsons and Mr. S. S. Cook .
Summary of Experimental Results .
( 1 ) The energy emitted in the form of Rontgen radiation by a cathode particle when suddenly stopped is proportional to the fourth power of its velocity .
( 2 ) The primary rays from a Rontgen ray tube can only excite the radiation characteristic of a radiator of atomic weight when the velocity of the parent cathode rays exceeds cm .
This law holds fairly closely for the radiators Al , Cr , Fe , Ni , Cu , and Se .
It gives me pleasure to thank Prof. Sir J. J. Thomson for the interest he has taken in these experiments .
Experiments on tloe Compression of Liquids at High Pressures .
By the Hon. C. A. PARSONS , C.B. , F.R.S. , and S. S. COOK , B.A. Cantab .
( Received May 10 , \mdash ; Read May 25 , 1911 .
) 1 .
Introduct During the experiments on the behaviour of carbon under high pressures and temperatures , of which an account was given in the paper read before this Society by the Hon. C. A Parsons on June 27 , 1907 , very considerable volumetric compressions were observed , and the apparatus then employed appeared to be equally suitable for the direct measurement of the compressibility of liquids at higher pressures than had previously been attempted .
The measurement of the compressibility of liquids has received the attention of a great many investigators ; the most comprehensive researches in this subject appear to be those of Amagat , who determined the coefficients of compressibility of water and ether for )ressures up to 3000 atmospheres and for a variety of temperatures .
Amagat 's experiments are given in various numbers of ptes Rendu In the experiments about to be described , the were carried up to about 40 tons per square inch , or over 6000 atmospheres .
'Comptes Rendus , ' vol. 103 , p. 429 ; vol. 105 , p. 1120 ; vol. 10 ; vol. 108 , p. 228 ; vol. 111 , p. 871 .
1911 .
] On the Compression of Liquids High Pressures .
333 The experiments were commenced in 1908 , but , the first experiments showing some modification of the apparatus to be desirable , principally with a view to making temperature measurements , it is the results of the later experiments carried out during the last year that are chiefly brief reference is desirable , however , to the preliminary readings obtained , by way of explaining the method adopted to eliminate friction , and in order to indicate how these experiments suggested an extension of the research to the investigation of the effects of internal molecular forces of the liquids tested .
2 .
Apparatus .
The apparatus consisted of a steel mould of about 4 inches bore and 12 inches external diameter , placed under a heavy hydraulic press , capable of exerting a pressure of 2000 tons , with a main ram of 29 inches diameter and a 6-inch lifting ram .
In the first series of these experiments the mould was constructed of gun steel , having an elastic limit of about 40 tons per square inch .
Greater strength could have been obtained , and still higher pressures applied , by the employment of special steel with a higher elastic limit .
Water was supplied to the upper side of the ram at pressures up to 2000 lbs. per square inch by a three-throw hydraulic pump driven by an electric motor .
The pressure on the ram was recorded by a carefully Kialibrated Bourdon gauge , and the depression of the liquid in the mould measured by a pair of multiplying callipers , inserted between the top of the mould and a projecting collar on the plunger .
One of the outer arms of the callipers moved over a graduated scale attached to the other arm , and the readings of this scale were for safety observed through a telescope outside the armoured building in which the press and mould were set up .
A sketch of the apparatus is given in , fig. I. Between the plunger and the liquid was placed a leather cup backed by a thin-edged cup of brass , this combination making an effective packing ( see fig. II ) .
3 .
Elimination of Errors .
In such an apparatus certain errors would arise in a experiment , but the experiments were carried out in such a manner as to practically eliminate them .
The sources of error were the friction of the , the compression of the cup leather and brass and of the plunger below the point of measurement , and the lateral expansion of the mould .
All these , with the exception of that due to friction , would be eliminated if an experiment could first be made with the mould filled with an incompressible fluid , and the apparent 334 Hon. C. A. DIAGRAM I. compressions so obtained deducted meant .
But as there was no order to obtain in an indirect manner apparatus of a substance of negligib 1911 .
] On the Compression of Liquids at High Pressures .
335 cedure was adopted .
A steel cylinder was prepared of half the volume of the liquid and of smaller diameter than the bore of the mould .
In a first experiment this steel cylinder was inserted in the mould , which was then filled with distilled water up to a mark corresponding to a total volume of 2000 , so that it then contained 1000 .
of water and an immersed volume of 1000 .
of steel , both of which could be subjected together to bulk compression , and the apparent compression observed .
A second experiment was then made with the steel cylinder removed , and the mould filled to the same mark with 2000 .
of distilled water , the apparent compression being read as before .
Since the volumetric compression of the water in the first experiment is only one.half of its value in the second ( the original volume of the water compressed being only one-half ) , by subtracting the compression readings of the second experiment from twice those of the first , the compression of the water is eliminated , and the differe11ce thus obtained gives the constant error of the apparatus on the supposition that the inserted steel cylinder is incompressible .
A small correction is then necessary for the compression of this steel cylinder .
The coefficient of compressibility of steel at ordinary pressures , as deduced from modulus of rigidity and Young 's modulus , by the usual formula for the relation between the moduli , ' per atmosphere of pressure .
Since this is only a small percentage of the compressibility determined below for water , etc. , the reduction in volume of the steel .
cylinder under a pressure of atmospheres may be assumed , to a near enough approximation , to be cp times its original volume , where .
In the compressibility of water by the difference between the apparent compressions obtained in the two experiments , it is obvious that it is greater by approximately this amount ( viz. , than it would be on the assumption of absolute constancy of volume of the immersed steel cylinder .
A further small correction has been necessary owing to the expansion of the mould taking effect upon htly different volumes in different experiments .
This correction has been estimated by the usual theory of the bursting strains of thick cylinders , its magnitude at the highest pressures employed never exceeding 1 per cent. 4 .
Elimination of the Bffects of Frietion .
Curve 1 gives the readings without corrections obtained with the mould containing 2000 .
of distilled water at C. The vertical ordinate * See Thomson and Tait 's ' Natural Philosophy , ' S683 .
1911 .
] On the Compression of Liquids at High Pressures .
337 represents the volume of the liquid as indicated by the depression of the plunger , and the abscissa the pressure in atmospheres .
The cyclic nature of the curve will be noticed , the upper branch of the loop being plotted from the readings obtained when the pressure was increasing , and the lower from those obtained whilst the pressure was being reduced ; the difference between the readings of the upper and lower branches is concluded to be due to friction , the horizontal breadth of the loop being twice the pressure needed to overcome the friction of the packing and of the working parts of the press .
Loops of this nature were traced for all the experiments .
To guard ainst leakage , after the pressure had been raised to its highest value and reduced back to zero , it was again raised so as to repeat a portion of the higher branch of the loop ; if any leakage had taken place it was thus at once detected , and the experiment ected .
By drawing a curve midway between the two branches of the loop we obtain a curve of compression with the effects of friction eliminated .
Curve 2 is the curve plotted for 2000 .
of water in this manner .
Curve 3 is a curve of errors of the apparatus obtained by a calibrating experiment with the immersed steel cylinder as described above .
Curve 4 gives readings plotted for a heavy cylinder oil of mean density about at C. It will be noticed that the loop of Curve 4 is very similar to that for water , Curve 1 , in spite of the greater viscosity of the oil , so that it would appear that the loop obtained in all the experiments with liquids is entirely due to friction of the press and packing , and not to internal friction .
5 .
of Graphite under Pressure .
Similar measurements were attempted with powdered Atchison graphite of specific gravity and containing per cent. of ash .
The obtained are plotted in Curve 5 .
It was thought possible that under high pressure graphite would behave as a fluid , but the character of the curve obtained indicates an increasing degree of consolidation as the pressure is increased .
The consolidated graphite appears to show considerable rigidity , that is to say , the stress was not equal in all directions .
The lower portion of the cycle was repeated , applying and removing the pressure a great number of times , the dotted Curve 6 representing the readings subsequently obtained .
On opening up the mould the graphite was removed in biscuit-like ' fragments , often separating along a conical surface at about to the axis , so that it appears to have relieved itself during expansion along lines of cleavage at this angle .
Similar results were obtained with Ceylon graphite of specific gravity and containing about 15 per cent. of ash .
The readings for this graphite au plotted in Curve 8 .
6 .
Effects of Heat of Compression .
It will be observed that in the curves plotted for the experimental readingS , the readings for the pressures ascending a second time for the purpose , as explained above , of detecting leakage , are in general very slightly below the readings for the previous application of the pressure .
Subsequent examination of the apparatus showing no trace of leakage , and this phenomenon persisting in subsequent experiments , it was concluded to be due to temperature variation , the work done during compression transformin heat and rendering the substance , whilst under high pressures , slightly hotter than its surroundings , so that at the end of the cycle it had lost some heat to the surroundings , and during the second compression was slightly colder than during the first .
This heat effect being therefore a possible source of error , it was desirable to make some measurement of its amount .
As a preliminary , a volume of 2000 .
of heavy cylinder oil was pressed to a pressure of 6300 atmospheres , which was maintained for two hours to allow the heat due to compression to leak away .
At the end of this time the pressure was suddenly released , and the liquid was found to be 1/ 5 inch lower in the mould than it was before compression , corresponding to a reduction in total.volume of about 2 per cent. The heat effect was further investigated by allowing 2000 .
of thin machine oil to cool under high pressure for hours with an external temperature of about C. , and after sudden removal of the pressure opening up the mould for examination .
With the method of packing employed at this time it was not possible to open up in less than about a quarter of a hour , owing to the difficulty of removing brass cup and the cup-leather from the mould .
The temperature after up was read , however , at regular intervals of time , by which means , a logarithmic law , the temperature immediately after release and expansion could be estimated , and appeared at the moment of release to have been about C. It was evident from this that the loss of heat was considerably in excess of 1911 .
] On the Compression of Liquids at High Pressures .
341 about 41bs .
, which is equivalent only to heat units per lb. , leaving units per lb. to be accounted for by internal forces .
This point having been reached , the experiments were continued with a twofold object , namely , ( 1 ) to ascertain the compressibilities of various liquids for both adiabatic and isothermal compression , the experiments last referred to showing it to be necessary to discriminate between these modes of compression , and ( 2 ) to determine the amount of the heat developed from internal forces during compression and absorbed by the same agency during .expansion .
Various attempts were made to measure the instantaneous temperature of the contents of the mould during compression by electrical methods with platinum coils and terminals brought outside the mould to a galvanometer .
None of these were successful ; the oeadings interference from galyanic action and from conduction in the liquid , and constant difficulty was experienced through of the coils with congealed oil and through leakage of the fluid past the terminals , in spite of many precautions taken to avoid these difficulties .
In the meanwhile an accumulator had been installed , and it was now possible to put on or remove the full pressure almost instantaneously ; further , by attaching the packing to the ram in such a way that it could be withdrawn mmediately on the removal of pressure ( see Diagram I , fig. III ) , it was possible to insert a quick-reading mercury thermometer immediately after release .
The mould was also surrounded by a water jacket so that the temperature could be varied as required and measured by a mercury thermometer .
Isothermal curves were plotted by allowing the pressure to remain applied long enough for the liquid in the mould to assume the temperature of the jacket , after which the pressure was varied slightly above and slightly below its recorded value sufficiently to overcome the friction .
For adiabatic compression the pressure was raised to the required value in a few seconds .
In these later experiments the ressures were not carried higher than from 4500 to 5000 atmospheres , the adoption of which lower values , together with the experience acquired in with these high pressures , rendered it safe for the observer to work inside the building close to the apparatus , which greatly facilitated the readings .
7 .
Bxperiments with Wate Curve 9 is an isothermal curve obtained for the compression of distilled water at C. plotted to a base of pressure in atmospheres , the vertical ordinates giving the volumes of a kilogramme in cubic centimetres .
In the case of water the adiabatic compression was found to be only about VOL. LXXXV.\mdash ; A. 2 3 per cent. less than the isothermal .
With other substances , as will be below , the difference was greater , on account of lower specific heat and apparently higher internal forces .
Curve 10 is a curve of adiabatic expansion of water from a temperaturg of C. at 4550 atmospheres pressure .
The temperature at the end of this cent. lesstetHon .
C. A.Parsons and Mr. Cook .
, as [ May process of adiabatic expansion was found to be C. , so that 1 calories per kilogramme have to ) restored to bring the water back to C. at atmospheric pressure .
The external work done during expansion is given by the shaded area under the curve and amounts to 2700 per kilogramme .
But the 13 calories of heat which have to be added per ramme are equivalent to 5500 kilogramme-metres , so that in passing from the initial state of C. temperature and 4550 atmospheres pressure to a final state at the same temperature and atmospheric pressure 2800 kilogramme-metres of energy have become latent , or in other words have been converted into internal potential energy .
The increase of volume between these two conditions is 130 , and this disappearance of energy is therefore equivalent to the work that would be done against an internal force of average value , or 2150 atmospheres .
1911 .
] On the Compression of Liquids at High Pressures .
343 It will be seen that by increasing the pressure from 1 up to 4500 atmospheres water at C. is compressed to 87 per cent. of its original volume .
At 3000 atmospheres it is per cent. This agrees closely with per cent. given by Amagat for 3000 atmospheres pressure at C. An endeayour was also made in the case of water to find the influence of temperature upon compressibility .
For this purpose the water was compressed adiabatically at various temperatures .
Readings were taken of the compression at 600 atmospheres and immediately afterwards at spheres , the interval elapsing being only a few seconds .
The amount of the compression between these pressures is plotted in Curve 11 , the units adopted being of volume in cubic centimetres per kilogramme for the vertical ordinate and temperature before compression for the horizontal base .
The compressibility is reduced with increase of temperature , from which it follows that the coefficient of heat expansion at high pressures is htly greater than at atmospheric pressure .
This is in agreement with Amagat 's results .
8 .
Experiments with Ether .
Curve 12 shows the compressibility of pure ether at C. , the vertical ordinate being the volume iu cubic centimetres of a quantity whose volume is 1000 .
at C. The compression for 4000 atmospheres is from .
to 840 c.c. , the final volume being 80 per cent. of the , so that the compression in this case is practically double that obtained for water .
The agreement with Amagat 's results in this case also is fairly good .
the latter giving a compression from 1050 to 863 for 3000 atmospheres at C. , whilst by the present experiments it appears to be from 1050 to 870 , the disagreement being only about 4 per cent. of the total compression measured .
Hon. C. A. Parsons and Mr. S. S. Cook .
[ May 10 , Neither for ether nor for water can the curves showing the results of these experiments be said to definitely indicate the existence of a limiting value to the compression .
In the case of ether , as in the case of water , experiments were made to ascertain the heat absorbed through the action of intermolecular forces by sudden expansion .
Curve 13 is the curve of adiabatic expansion after cooling at 4400 atmospheres to C. The final temperature was C. This drop of C. , assuming the specific heat of ether to be corresponds Hon. C. A. Parsons and Mr. S. S. Cook .
[ May 10 , to 871 , or about 84S per cent. of its original volume at C. ; the compressibility is therefore greater than for water by about 20 per cent. The , drop in temperature on expansion from 4600 atmospheres to atmospheric pressure was found to be C. Taking the specific heat at ) energy extracted per 1000 .
is made up of 2055 kilogramme-metres of external work and 4645 kilogramme-metres work against internal forces , the latter corresponding to a pressure of 2920 atmospheres .
10 .
Conclusion .
To exhibit the comparison between them , the curves of isothermal compression have been replotted in Diagram II , reduced to a common basis of 1000 .
original volume .
Expressing compressibility in atmospheric units , that is to say , as the ratio of the decrease of volume per atmosphere of pressure to the volume of the liquid , from the foregoing experiments the following values have been deduced for the isothermal coefficients of compressibility of water , ether , and paraffin oil aterEther The average values of the molecular force deduced from the heat lost during adiabatic expansion as described above are as follows.\mdash ; Molecular force .
atmos .
Water ( for pressures between and 4550 atmos .
) 2150 Ether , , , , , , 4000 , , 2440 Paraffin , ; , , , , 4600 , , 2920 The authors wish to record their thanks to Mr. Robert Howe with the mechanical work in connection with the experiments .
1911 .
] On the Compression of Liquids at High Pressures .
.347 ( June 3 ) .
The method adopted in the paper for determining the equivalent internal pressure corresponding to the excess of heat extracted over the external work done is based on the following considerations:\mdash ; Suppose a closed cycle on a pressure volume diagram , made up of the following processes in isothermal compression at temperature ; ( b ) adiabatic expansion to atmospheric pressure and temperature ; ( c ) restoration to temperature by addition of heat at atmospheric pressure .
The external work during process ) is negligible .
Hence the excess of the heat given out over external work done during ( a ) can be measured by the excess of the heat absorbed during ( c ) over the external work done during The excess thus determined in has been expressed as the product of an average internal pressure into the change of volume occurring during the isothermal compression .
The rate of rise of temperature of fluids under compression can be determined from the thermodynamic relations resulting from the first and second laws of thermodynamics .
When there is no emission of heat , we have with the usual notation , ( 1 ) from which .
( 2 ) Using to denote the coefficient of expansion at constant pressure , or ' we have .
( 3 This formula will ordinarily only bear application in the neighbourhood of atmospheric pressure , the values of and under the high pressures employed in the foregoing experiments being unknown .
Thus , for the adiabatic curve plotted for water , in the neighbourhood of atmospheric pressure , we have for the various expressions in equation ) kgrm.-cm .
per kgrm . .
per kgrm .
from which there results , or a rise of C. per atmosphere of pressure .
The rise in temperature will , however , in the case of water , proceed more rapidly in the later stages of the compression as higher pressures reached , since increases both with the temperature and with the pressure .
On the Compression of Liquids at High Pressures .
With the help of Curve 11 it is possible to obtain an approximate estimate of the value of of the fluid when under compression .
Denoting by the compressibility at constant temperature , since and it follows that The slope of Curve 11 gives for the value of , with the unit adopted for ( viz. , change of volume per unit volume per atmosphere of pressure ) , , and if we assume this as the approximate rate of increase of throughout the range of pressures covered by the adiabatic compression from C. at atmospheric pressure to C. at 4550 atmospheres pressure , we have for the value of at the beginning of the compression , and at the end , or an average value of The expressions in equation ( 3 ) then assume the following average values over the whole range of this compression\mdash ; .
per kgrm .
, giving per atmosphere .
The measured drop in temperature during the reverse process of adiabatic expansion , from a pressure of atmospheres at C. to atmospheric pressure , was C. , or per atmosphere , thus agreeing even closely with the estimate .

