RA'IMO OF SPECIFIC HEATS. 9 



positive or negative according as tlie resistance in Z Ls lai'ger or smaller tluin the 

 equilibriiun vahir. Wlicii the ilctlectidii ilue to the changed resistance in Z lias 

 become permanent, let the receiver be opened and note the effect of expansion 

 on the galvanometer needle, 'ilicii according as the arbitrary change of resist- 

 ance at Z is too large or too small, the needle will move beyond its normal posi- 

 tion of eqiiilibi'inm, or it will stop and return before reaching it from the new 

 position of equilibrium. ( )\ving to the inertia of the needle such I'esultsare not very 

 accurate, and they are in laige measure dependent on the rate of efHux of the 

 gas. A rather insensitive galvanometer (or one made so by [)roper shunting) is 

 necessarv in these ti'ials Itecause of the relatively great resistance decrements 

 encountered. 



If it be granted that the temperature of the gas is always identical with that 

 of the bolometer strip, and that furthermore the maximum of cooling coincides in 

 date \vith the maximum elongiition of the needle, then in addition to furnishing 

 approximate values for w^-iu.;^ and Pi-^^g, the experiments throw light also on the 

 time consumed by the gas in passing to its lowest temperature. 



Guided by this introductory work we were able to avail ourselves of the fol- 

 lowiui;- much more accurate method without unduly sacrificing time. Let the 

 pressure |>] (known approximately from the preceding experiments) be established, 

 and let the bridge be adjusted for the temperature J\ . The approximate value 

 to^-w., corresponding io p^ being given, let the current be broken and the resistance 

 of the standards Z be lowei'ed from W^ to W», without change in the position of 

 the sliding contact c. While the bridge is thus thrown out of adjustment and the 

 current broken, the gas is allowed to expand, but the current ' is to be made by 

 closing the key n of the bridge only after the bolometer has reached its mini- 

 mum resistance. 



After some practice it was always possible to so change the initial pressure by 

 small amounts and so vary the estimated period of expansion, that the needle of 

 the galvanometer remained quite at rest- for some time after making the current. 



When the correct pressure p^ has been found, this value may easily be checked, 

 since for initial pressures either slightly less or slightly greater than the true value 

 the galvanometer will show positive or negative deflections, respectively ; and thus 

 the datum jh can be enclosed between very narrow limits. 



' It was found preferable to close the current in the battery circuit, not in the galvanometer cir- 

 cuit, in order to obviate errors arising from the presence of thermoelectric currents in the bridge. 

 No 'temporary thermocurrents arise during the efflux of the gas, as we proved by special efflux 

 experiments made for a circuit open at n. The galvanometer needle remained quite at rest. 



' We owe this method (" Einspringmethode ") to Dr. Kurlbaum, who took part with us in the 

 original experiments in 1885. 



