M. Mcndclejeff on the Compounds of Alcohol icith Water. 137 



By substituting these values, the formula gives 



J = 422-10±O57; 



if the probable error of the velocity of sound alone is considered, 



log J = 2-62542. 



Considering this velocity, again, correct to To Vo- the value of 

 J must be between 423*82 and 420-38. All the other magni- 

 tudes used in the calculation could not possibly produce a greater 

 error : the greatest error might still arise from the value of c„ 

 although this passes in a lessened form into the value of c l — c. 

 The calculation has been given completely, in order the more 

 readily to calculate the corrections which J would experience 

 from a better determination of the elements. The statement of 

 the limits within which the value of J may be taken as being 

 tolerably certain was made in order, in the first place, to esti- 

 mate the degree of trustworthiness which this determination can 

 acquire, and, in the second place, because it leads to far more 

 precise results if other determinations, such as those of Clausius 

 and of Joule, undergo the same treatment. It is remarkable 

 that the determinations, of Joule = 42355, of Clausius =421, 

 and of Bosscha, developed from galvanic measurements, =421*1 

 (Pogg. Ann. vol. cviii. p. 169), all fall within these limits; and 

 the value which has been found is almost exactly the middle 

 of these. Hence in the subsequent calculation I have adhered to 

 this value, and not combined it with other results of whose degree 

 of trustworthiness I could not judge so well. 



[To he continued.] 



XVIII. Chemical Notices from Foreign Journals. 

 By E. Atkinson, Ph.D., F.C.S. 

 [Continued from vol. xxx. p. 451.] 



THE Zeitschrift filr Chemie for May, 1865, contains the fol- 

 lowing abstract of a paper by Mendelejeff, in the Russian 

 language, on the compounds of alcohol with water. 



The first chapter the author devotes to a detailed and thorough 

 criticism of all previous researches on this subject. He deter- 

 mines the magnitude of the errors of observation of the different 

 authors, reduces all their weighings, and endeavours to make all 

 determinations comparable by reducing them to the same unit. 

 The author comes to the conclusion that the old observations of 

 Gilpin (1794) are among the most accurate. With these the 

 observations of Fownes (1847) and of Drinkwater (1848) closely 

 agree. The numbers of Gay-Lussac and of Trallcs cannot stand 



