28 Dr. E. J. Mills's Researches on Melting-point. 



final results. It is to comparison with the air-thermometer 

 that we have to look for any important source of error. All 

 observers who have made an extended range of such compari- 

 sons have found noteworthy errors, though they have in no 

 case stated probable error. The probable error of the result 

 of my own comparisons of thermometer 2 with the air-ther- 

 mometer is 0°*085 for a single set of comparisons, or o, 085 

 -r-v / o , 3 = o -015 for the results of the thirty-three sets. This 

 number is the measure of probable error of the equation em- 

 ployed in the final reductions. Compounding, then, ihe mean 

 probable error of the melting-point (-004) with that of the 

 comparison ('015) with the air-thermometer, we may consider 

 the melting-points in Table XXV. ascertained, in terms of 

 the air-thermometer, with a probable error of 



-v/(-004) 2 + ('015) 2 = 0-015. 



The relation of the chemical symbol to the physical proper- 

 ties of a substance is a matter of such great interest that I have 

 sought for it in melting-point, although other investigators of 

 the general subject, working with less definite data, have not 

 arrived at very encouraging results. 



It is very easy to show that, in some cases, there is a very 

 simple connexion between the formula and the melting-point 

 of a substance in the centigrade scale. Thus, dichlorobenzol, 

 bromaniline, and trinitrotoluol form a group in which melting- 

 point = <j> x numerical value of formula. 



Substance. Formula. Melting-point. cf>. 



Dichlorobenzol . C 6 H 4 C1 2 =147 52°-821 "35933 



Bromaniline. . C 6 H 6 BrN=172 61*742 -35971 



Trinitrotoluol . C 7 H 5 N 3 6 =227 80-532 -35477 



In the first of these two instances the values of (f> are almost 



exactly the same ; in the last, however, the limits of probable 



error are exceeded, though a close approximation is very 



evident. 



The following comparison furnishes another practical iden- 

 tity:— 



Melting-point. Melting-point. 



Trinitrotoluol. 78-853 - Dinitrotoluol . 69-252 = 9-601 \ 

 Trinitrophenol . 121-194 — Dinitrophenol. 111-621=9-573/ 



In the next instance there is an approximation : — 

 Melting-point. Melting-point. 



Dinitrotoluol . 69*252 — Nitrotoluol . 51-407= 17*845 \ 

 Dinitrophenol. 61*843 — Nitrophenol. 44*392 = 17*451 J 



