254 Professor Sir James Dewar [Jan. 23, 



Anhydrous glycerin was used to make the ground joint tight at the 

 low temperatures. The wider neck in some ways simplifies the 

 preliminary manipulations, and allows a more efficient arrangement 

 to be securely fixed for preventing splashing, and breaking the fall 

 of the bodies dropped into the calorimeter. The arrangement is 

 shown in the side sketch B (Fig. (J). It consists of a light counter- 

 poised trap door hinged at the lower end of a conical brass or german 

 silver tube, supj^orted in and fitting to the inner tube of the calori- 

 meter. A small lead ball on a wire soldered to the trap door keeps 

 this closed until struck by a falling sphere, while shutting it again 

 immediately after the sphere passes through. The conical tube above 

 the trap door is pierced with several small holes to leave a free passage 

 for the evaporating hydrogen. The addition of a gauze filter to inter- 

 cept spray was found to be impracticable, the resistance introduced 

 causing back pressure. In later work the calorimeters were con- 

 structed of quartz, a little charcoal in the bottom of the annular 

 space serving to maintain a very high vacuum isolation. 



As far as possible the materials used were cast in the forms of 

 spheres about 8 mm. diameter, and for this purpose the use of an 

 ordinary bullet mould was found convenient. In the case of liquid 

 bodies, the mould was first cooled by liquid air. Frequently liquids 

 were frozen into solid cylinders in thin glass tubing, and pieces cut 

 off after removing the glass mould. The metallic materials were in 

 some cases fused into buttons of convenient weight in an exhausted 

 quartz tube. The lead, however, of which many pieces were re- 

 quired, were cut from rod, and subsequently squeezed in a small 

 spherical mould. 



Volatile bodies were Aveighed at a low temperature on a light 

 german silver pan supported by a thin platinum wire suspension 

 from the balance pan about 2 cm. above the level of liquid air 

 contained in a wide deep vacuum vessel. Some materials would not 

 make coherent bullets or cast sticks. These were filled into very 

 thin walled cylindrical metal capsules of equal weight, so that a 

 preliminary determination of the volume of hydrogen evaporated by 

 the metal of the capsule gave the correction. They were then cooled 

 on an aluminium dish floating on Hquid air, filled with the fluid, and 

 weighed separately. Materials which could not be fused were com- 

 pressed hydraulically into small blocks and cut up into pieces of 

 suitable dimensions. 



At least three pieces of every substance were dropped. The 

 results rarely varied among themselves by more than 2 to 3 per cent. 

 Very frequently the agreement was within 1 per cent. In order to 

 ensure good results, uniformity of shape and size in the pieces of 

 material used is desirable, so that the manner of release and fall 

 shall be comparable : because in the use of this instrument the 

 materials had to pass through a region of the neck, between the 

 cooling vessel and the calorimeter, where a considerable gradient of 



