OF ARTS AND SCIENCES. 239 



mercurial thermometer at those temperatures. For benzophenone 

 (p. 247) the accordance is from 0°.3 to 0°.5 C, — a difference possibly 

 arising from errors in thermometry introduced by difficulties met with 

 on our part from somewhat irregular action of the kind of glass which 

 we were forced to employ in our air thermometer bulb, and which 

 rendered the determination of its coefficient of expansion somewhat 

 less satisfactory at this higher temperature than at lower ones. Yet 

 we think the difference too great to be wholly accounted for by 

 thermometric errors. 



The results of the investigation may be summarized as follows : — 



1. Naphthaline, C 10 H 8 , is readily obtainable in a state of sufficient 

 purity to give a reference temperature exact within 0°.l C. We have 

 found a preparation melting at 79°.4 to 79°. 8 to possess a boiling point 

 within the ordinary range of atmospheric pressure {H), expressible by 



f = 218.07 + 0.0625 (H - 760), 



where -ff"is the reduced pressure in "normal " millimeters of mercury. 

 Throughout the paper the degrees and pressures may be regarded as 

 "normal," i. e. corresponding to the definitions adopted by the Inter- 

 national Committee of Weights and Measures, Trav. et Mem., i. (1881). 

 No reduction for gravity has been made, because the correction at 

 Boston is below the limits of error of this work. . 



2. Benzophenone, (C 6 H 5 ) 2 CO, is obtainable with some difficulty, 

 and is rather costly. With a melting point of 47°. 6 to 48°.0 our de- 

 termination of the boiling point within the ordinary range of atmos-. 

 pheric pressures is expressible by 



f = 305.6 + 0.060 (H- 760). 



3. Benzol, C 6 H 6 , is readily obtainable nearly pure. Anhydrous 

 benzol melting at 4°. 22 was found to have a boiling point expressible 

 within the ordinary range of atmospheric pressures by 



f = 80.19 + 0.0455 (H- 760). 



4. The boiling points for pressures down to 80 mm. for naphthaline 

 and benzophenone, and to 360 mm. for benzol, are tabulated on page 

 247. No attempt has been made to express the vapor pressure as a 

 function of the temperature through these greater ranges by any of 

 the numerous formulae employed by others for this purpose. 



5. We regard the actual errors in the final results for the naphtha^ 

 line and benzol at the atmospheric pressure as under 0°.l, and that 

 for the benzophenone as under 0°.25. The average deviation of the 



