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



Table XXV. 

 Summary of Results. 



27 



Substance. 



Weigbted 

 mean. 



Probable 

 error. 



After 



PoggendorfFs 



correction. 



Air- 

 tbernionieter. 





42-765 

 44-270 

 51-305 

 52-723 

 56-175 

 61-778 

 61-806 

 69-211 

 69-571 

 69-667 

 70-598 

 77-052 

 78-821 

 78-841 

 80-061 

 80-524 

 83-490 

 87-037 

 89-718 

 111-413 

 111-579 

 116-247 

 121-082 



•004 

 •003 

 ■005 

 •002 

 •002 

 •003 

 •003 

 •004 

 •004 

 •003 

 •004 

 •001 

 •004 

 •004 

 •002 

 •003 

 •002 

 •002 

 •003 

 •006 

 •004 

 •005 

 •005 



42-700 

 44-205 

 51-239 

 52-657 

 56-110 

 61-714 

 61-742 

 69-154 

 69-514 

 69-610 

 70-542 

 77-004 

 78-776 

 78-796 

 80018 

 80-481 

 83-452 

 87-007 

 89-693 

 111-448 

 111-614 

 116-298 

 121-151 



42-890 

 44-392 

 51-407 

 52-821 

 56-261 

 61-843 

 61-871 

 69-252 

 69-610 

 69 706 

 70-634 

 77 068 

 78-833 

 78-853 

 80-070 

 80-532 

 83-492 

 87-035 

 89-712 

 111-455 

 111-621 

 116-319 

 121-194 









Nitronapbtbalin 



Dinitropbenol (a) 



Monobroruaniline 



Dinitrotoluol (a) 



.. „ (*) 



Monocbloraniline 



Dinitrobroruobenzol . . . 







Napbtbalin 



Trinitrotoluol (M) ... 

 Nitrodibromobenzol . . . 















Discussion. 



The determinations of melting-point which have been 

 recorded in the preceding tables, and the results of which are 

 summarized in Table XXV., show a very small probable 

 error in connexion with their weighted means. The probable 

 error of a weighted mean has ranged from o, 001 to o, 006, 

 its average value being less than o, 004. So far, then, as 

 regards the actual process of ascertaining melting-point, con- 

 siderable accuracy has doubtless been attained. 



The preliminary operation of calibrating the thermometers 

 was so conducted as not to have introduced material error, as 

 indeed is obvious from a comparison of the results obtained 

 with different thermometers on melting the same substance*. 

 Regnault was of opinion that the height of the barometer 

 cannot be ascertained with a less error than about 0"1 millim. 

 Such an error would correspond to about 0°'0037 on 100°; 

 and the error would be still less on the mean, as in the present 

 case, of several readings. The small errors in the determina- 

 tion of the exposure-corrections could not sensibly affect the 

 * See, for instance, Table VII. 



