HEAT PRODUCTION BY EGGS. 343 



means of the automatic intermittent siphon device shown in 

 Fig. 2. A steady stream of water flowed in at (a) and as the 

 water level rose in the bottle, air was trapped in (b, b) and 

 forced out through (d, d}. When the water reached the lower 

 end of (e) it rose in the tubes, but not in the bottle (since the 

 stopper was air tight) until ' (/) was filled beyond the bend. 

 Then (/) acted as a siphon and emptied the bottle, allowing 

 (b, b) to drain, and finally draining itself. Then the process 

 started over again. Short pieces of rubber tubing cut at an 

 angle were slipped over both ends of (/) and the lower ends of 

 (b, b) to help in breaking the meniscus and draining these tubes. 

 The quantity of air sent over each time could be varied by 

 adjusting the height of the small tubes (c, c) in (b, 6), and the 

 period of the apparatus could be changed by raising or lowering 

 (e). The small bulbs (d, d) prevented drops of water from 

 being forced over into the wash bottles. These wash bottles 

 had inlet tubes of small cross section, and were submerged in 

 the same bath with the flasks. They served to complete the 

 saturation of the air, if it were not already saturated, to bring 

 it to the temperature of the bath before it went into the flasks 

 through (c, c), and to act as a trap, preventing the suction of 

 liquid out of the flasks through (c, c) when the stirrer bottle 

 was emptying. In this work 7 cc. of air was sent through each 

 flask every fourteen seconds. During a six-day continuous run, 

 this apparatus did not vary from its average period by more 

 than 2/5 second, always delivering the same and equal quantities 

 of air to the flasks. 



ELECTRICAL EQUIPMENT. 



The electrical equipment used in these experiments consisted 

 of the main and secondary thermopiles, galvanometer, potenti- 

 ometer, storage battery cells, Weston Standard Cell, switches, 

 etc. The main thermopile (a, Fig. i) was composed of twenty 

 pairs of copper-constantan junctions, and gave 733 microvolts 

 per degree Centigrade, by direct calibration. (The constantan 

 wire used was that manufactured by the Driver-Harris Wire 

 Company, and sold under the trade name "Advance.") The 

 reason for the unexpectedly low value of the thermopile was 

 not found. The junctions were enclosed in thin glass tubes 



