OP ARTS AND SCIENCES. 07 



obviate the possible electric contact between W and ZZ, the hard 

 rubber jacket T is suitably flanged. XX is also to be flanged (not 

 shown). 



9. Piezometer Tube and Vapor Bath. — The outer end of UUU'is 

 closed by the tinned screw V. The substance to be examined is 

 here introduced, usually in glass tubes ; but the special devices to 

 be adopted for each special experiment are beyond the scope of the 

 present paper. As a rule, the glass sample tubes are adjusted within 

 the compass of the copper vapor hixih. ffff {¥'\g. 5), which surrounds 

 the steel tube UU eccentrically. The vapor bath is merely a long 

 cylinder of brazed sheet copper, containing the liquid yy, to be boiled 

 by heating it with a large burner. Vapors escaping at g are condensed 

 (condenser not shown) and run back, thus making the ebullition con- 

 tinuous. A number of baths of this kind are at hand, each containing 

 a substance of suitable boiling point. In order to pass from one tem- 

 perature to another, it is merely necessary to slip oflT one vapor bath, j^, 

 and slide on another. The copper vessels j^ are of course surrounded 

 with thick asbestos jackets, and proper provision is made to close up the 

 ends of the end tubes hh. As far as temperatures not exceeding 200°, 

 perforated rubber corks are satisfactory. For higher temperatures, hh 

 are stuffing boxes containing asbestos wicking tightly appressed. It is 

 preferable to lead off the vapor by a long lateral oblique gas-pipe (not 

 shown) and to use the tubulure g for the insertion of the thermo- 

 couple * for measuring temperature. A tube or jacket through which 

 cold water circulates must surround the piezometer UU, between ff 

 and the barrel, and between ff and the end screw F, otherwise the 

 heat conducted along the tube UU will eventually melt both the 

 marine glue and the solder joints. It is for the same reason (heat 

 conduction) that annular vapor baths of the kind described in the 

 Bulletin of the U. S. Geological Survey, No. 54, Chapter IT., are not 

 admissible. Ileat is conducted along the metallic tube more rapidly 

 than it is supplied, and constancy of temperature cannot be guaran- 

 teed nor easily measured. It is essential that the vapor be directly 

 in contact with the piezometer tube. Indeed, when very fine tem- 

 perature work below 200° is necessary, I use the drum ff simply for 



sary. I may add, that as an insulator marine ghie is not absolutely perfect, tbe 

 resistance here being a few hundred thousand ohms. For very fine electrical 

 work some other material may be used. 



* Cf. Barus, Bull. U. S. G. S., No. 54, 1889, p. 22 ; Phil. Mag., Vol. XXIX. 

 p. 141, 1890. A special torsion galvanometer, used for measuring the iliermo- 

 electric powers, is to be described elsewhere. 



VOL. XXT. (N. S. XVII.) 7 



