22 PROCEEDINGS OP THE AMERICAN ACADEMY 



rignac. Of all the ratios between the atomic weights, it is the one in 

 regard to which there is the greatest certainty; and it is with this very 

 well established relation that we have connected the atoir'C weight of 

 antimony. 



Entirely in harmony with the above results are our experiments on 

 the synthesis of sulphide of antimony, in which we found as a mean of 

 thirteen experiments Sb= 119.94 when S = 32 ; and the same is 

 equally true of our analyses of iodide of antimony, which gave as a 

 mean of seven determinations Sb = 119.98 when I = 127.00. But 

 although these results formed important stages in our investigation, 

 they now add but little to the evidence of the far more accordant re- 

 sults since obtained. When compared with these later results they 

 show, however, to what a great extent error may be eliminated by the 

 repetitions of an imperfect process. 



Lastly the anomaly which the analysis of antimonious chloride first 

 presented has been explained by finding, first, that the material em- 

 ployed contained a constant amount of oxichloride ; and, secondly, that 

 the water used in washing the precipitate exerted a definite solvent 

 action on the chloride of silver estimated. 



Having thus solved the problem we undertook as flir as is at present 

 practicable, we must now take leave of the subject, regretting only 

 that our investigation should have been the occasion of any controversy. 

 In addition to the recognition we made in our previous paper, we 

 would here express our obligations to Mr. G. De N. Hough and 

 Mr. G. M. Hyarns, who have greatly aided us in the experimental 

 work during the latter part of this investigation. 



V. — The Boiling Point of Iodide of Antimony, and a New 

 FoKM OF Air Thermometer. 



In continuation of our investigations on the Haloid Compounds of 

 Antimony, the boiling point of antimonious iodide has been deter- 

 mined by Mr. W. Z. Bennett, at the time a student in this laboratory. 

 The observations were made with Regnault's air thermometer, but it 

 was found possible to simplify very greatly the details of the process 

 without seriously impairing the accuracy of the result. For tempera- 

 tures above the range of a mercury thermometer's measurements, ac- 

 curate to one degree centigrade, are all that the uncertain conditions 

 of most problems permit, and all, therefore, that the circumstances 

 demand. As used by Regnault, the air thermometer is capable of 

 measuring such temperatures accurately to the one tenth of a degree, 



