On the Limit of the Liquid State. 



317 



middle part of the curve evidently reading a little low. The whole 

 result agrees well with what was shown from alcohol. 



The next body examined was methyl alcohol, a sample of which 

 was carefully purified in the same manner as the ethyl alcohol, until 

 its boiling point was constant. It was then distilled off quick-lime 

 into the wash-bottle arrangement for use. It gave as follows : — 



Mean temperature corrected 232 0, 76. 



Probable error o, 21. 



Mean pressure corrected 72*85 atmos. 



Probable error 0*12 „ 



These means were taken from three series, each of thirty experi- 

 ments. The same experiments were then carried out with methyl 

 alcohol and hydrogen and nitrogen, as in the case of ethyl alcohol and 

 carbon disulphide, yielding the following results from twenty-two 

 experiments : — 



Mean temperature corrected 230 o, 14. 



Probable error o, 09. 



Mean pressure corrected 128*60 atmos. 



Probable error 0*12 



Here we have, as before, a slight depression of the critical tempera- 

 ture. Experiments were then tried with higher pressure. Forty-seven 

 gave the following means : — 



Mean temperature corrected 227°*92. 



Probable error o, 10. 



Mean pressure corrected > 191 "40 atoms. 



Probable error 0'07 „ 



A still higher pressure was then applied. The means of eighteen 

 experiments were — 



Mean temperature corrected 225° *82. 



Probable error 0°"26. 



Mean pressure corrected 262*00 atmos. 



Probable error 0'09 „ 



We have a further depression of the critical point, and as the point 

 was determined with great difficulty we have an increase of the 

 probable error. However, the result confirms the other experiments. 

 The same mode of experiment was then carried out with methyl 

 alcohol and hydrogen and nitrogen as with ethyl alcohol, with the 

 following results. 



VOL. XXXIII. z 



