7 1 8 Journal of Agricultural Research vol. vi, no. is 



if the coils differ in resistance. Moreover, the ratio coils of one bridge 

 may be combined with those of either of the other three to form a test 

 bridge, all four arms of which should have the same resistance. With a 

 very sensitive galvanometer across the bridge thus formed, any inequality 

 in the coils would be detected. It is assumed that if there is no deflec- 

 tion the coils have not changed in resistance, since it is hardly probable 

 that all the coils would have changed equally. If any change should be 

 detected, by varying the combinations it would be possible to determine 

 which pair of coils was at fault. , By shifting the point of contact of the 

 battery lead on a short wire joining the two coils, equality of resistance 

 may be restored when the changes are slight. No tests of this character 

 have thus far indicated any need for change. Each bridge is also pro- 

 vided with a small variable shunt across a small resistance in series with 

 one of the two differential thermometers to compensate for small 

 inequalities in their resistances when at the same temperature. 



TESTS OF THE ACCURACY OF THE RESPIRATION CALORIMETER 



The accuracy with which it is possible under given conditions to meas- 

 ure the faqtors studied by means of the respiration calorimeter is shown 

 by a comparison of the determined amounts of oxygen consumed and of 

 carbon dioxid, water vapor, and heat produced upon combustion of ethyl 

 hydroxid in the chamber with those which should result from the com- 

 bustion as calculated from the quantity of alcohol burned and the per- 

 centage of ethyl hydroxid in it. 



A burner inside the chamber is connected with a small-bore copper 

 tube that passes through the "outlet" in the wall of the chamber. To 

 the exterior end of this tube is attached a glass U-tube with one long arm 

 into which alcohol for the burner is fed by dropping from a supply bottle 

 which may be weighed at intervals to determine the quantity burned. 

 To test the apparatus under conditions equivalent to those of experi- 

 ments in which it is used, the alcohol must be burned at a very slow rate. 

 Some difficulties were experienced at first in attempts to burn as little 

 as i gm. per hour with complete combustion of the alcohol at a constant 

 rate and with inappreciable loss by evaporation from the long arm of the 

 U-tube. These were due in part to the fact that the opening in the 

 "outlet" through which the alcohol tube passed is considerably above 

 the level at which it is desired to have the combustion take place in the 

 chamber. As a result of this condition, in all the tests thus far made the 

 level at which the alcohol was maintained in the vertical tube was above 

 that at which it was burned, attempts to feed the burner by siphon hav- 

 ing proved unsuccessful. It was necessary to devise a burner which 

 would overcome the effect of the pressure of the alcohol in the feed tube 

 upon the rate of flow. 



Burners of small-bore glass tubing of various diameters and with wicks 

 of cotton, of glass wool, and of ignited asbestos, packed so as to allow 



