A Comminication made to the TRANSACTIOXS of tlac Fnvaday Society.) A recent note on this subject by E. B. R. Prideaux :k induces me to ofkr 'two remarks ; they will, I think, be found useful for this kind of research. I. The formula of Meiidelejeff tas well as the formula of Thorpe and Riicker,; which follows from itare certainly very interesting and admit of being applied in such researches. It should never be overlooked, however, that these relations are merely approximations. In fact, the

more »

... mula of Mendelejeff may also be written- Thus developed into a series, the expression I/(Ik t ) only forms a special case of the usual empirical formula-I -k t Vt=Vo(l + at + biz + d 3 ...), but it offers less scope for determining the values arid the signs of tlie coefficients of the terms containing t ; it can, therefore, only give a smaller exactness . It is, moreover, easy to convince ourselves that when we calculate k for the same liquid, starting from diverse systems of values V" V" TI, T" the discrepancies will certainly much exceed the errors of observation. In accord with the theory experiments thus confirm that the relation (I) can only be approximately true ; it will in general be the more accurate, the less the dilatation of the liquid will deviate from the linear form. Somewhat the same statement may be made with regard to relation ( 2 ) , which-as is well known-results from a combination of formula (I) with a consequence of the theory of corresponding states. Xelation (2), however, or its equivalent-