34 SCIENCE PROGRESS 



lem awaits some one who is able to systematise the enormous 

 mass of data which is being obtained. Something in the 

 shape of a further generalisation has been obtained by the 

 application of the thermodynamic reasoning of J. Willard Gibbs 

 to the relations of the solid to the other states ; but this rather 

 extends the former results of van der Waals to states which he 

 did not consider, than increases the general accuracy with which 

 the experimental data are systematised and new relations deduced. 



One of the main difficulties in this subject is the great experi- 

 mental difficulty which is encountered directly really accurate 

 data at any other temperatures than the normal are required. 

 Even at the normal temperature it is only by the very greatest 

 care at every step that values are obtained, which are more 

 accurate than to 0*02 per cent. The vast majority of measure- 

 ments of compressibility at constant temperature, the deter- 

 mination of isothermals, are hardly accurate to o'2 per cent., while 

 very few critical data are accurate to i per cent. 



It is very rarely that the same observer makes measurements 

 on the three critical data, so that the results are often not very 

 comparable, and in any case the values given are in units which 

 are not always self-evident. It is unfortunate that a really 

 strict system of units has not been generally recognised, as all 

 three units of pressure, volume, and temperature are liable to 

 some ambiguity. Pressure is usually expressed in atmospheres, 

 the value of which depends upon the latitude of the experimental 

 station at which the determinations are made, but which are 

 sometimes mean atmospheres reduced to latitude 45 . If all 

 observers deduced their results to the C.G.S. unit of a mega- 

 dyne per sq. cm., which is very nearly an atmosphere, it would 

 be much clearer. The same is true of the volumes which are 

 sometimes given in the unit known as the normal volume, the 

 volume of the quantity of gas under experiment at zero C. and 

 under the unit of pressure employed. Others express the 

 volumes in terms of the mass of the gas, which is easily 

 converted to the first mentioned, if the law of Avogadro is 

 assumed to hold strictly. However, as will be explained later, 

 this law is not strict, and if a correction is applied so that 

 equal volumes of different gases shall contain equal numbers 

 of molecules, a unit is obtained which is known as the " theo- 

 retical normal volume " and which makes results on different 

 gases strictly comparable. There is less ambiguity about the 



