336 White — Specific Heats of Silicates and Platinum. 



tribution in the furnace and from the magnitude of the 

 accidental errors, and on that basis is almost certainly under 1 

 per cent at the extreme and 0*5 per cent at the intermediate tem- 

 peratures, and these values may be even smaller later on. 



2. The Drop. — Two features characterize the process of 

 dropping into the calorimeter and have rendered its results 

 very satisfactory. (1) The enclosing crucible is first tared as 

 to its thermal effect by dropping it alone, and the heat 

 quantity thus obtained is then subtracted from subsequent 

 determinations at the same temperature. This assumes that 

 practically all the heat loss in dropping comes from the cru- 

 cible alone, but as the time of fall is less than one-fourth of 

 a second, and the greatest total loss of heat is only 4'7 per cent 

 of that carried by the crucible, or 0*6 per cent of the total 

 quantity measured, this assumption seems justified. In 

 order to reduce, as far as possible, all variations in the heat 

 loss, the crucible, when dropped by itself, is loaded with 

 platinum so as to have about the same weight as at other times, 

 and hence to pass through the surface of the water at the same 

 speed. The specific heat of the platinum is so low that the 

 error in the correction for it is certain to be under 0*1 per cent.* 



(2) In the second place, the operations of dropping are per- 

 formed as quickly as possible. This is accomplished as follows : 

 (See fig. 1.) The bottom of the furnace is first dropped by pull- 

 ing a latch (L), and falls into a box (B), which is hinged so as to 

 swing horizontally out of the line of fall. A rapid glance at the 

 furnace bottom tells whether it has fallen properly, that is, with- 

 out the crucible, and with everything else that should come. 

 The box is then swung aside, and as it moves it automatically 

 shuts off the heating current of the furnace, and passes 40 

 amperes or so through a fine platinum wire by which the 

 crucible is suspended^ allowing it to drop. The immediate 

 return of the swinging box cuts off direct radiation from the 

 furnace to the calorimeter. The whole operation requires 

 about a second and the error from doubling the time, which 

 has been several times measured, is negligible. Schemes for 

 moving calorimeter or furnace have often been tried by others 

 and were contemplated here when the present work was 

 planned. But the device just described, which amounts to 

 moving only a light wooden shield between the two, was tried 

 first for its simplicity, and soon proved to be altogether the 

 most reliable and effective, reducing the time to a minimum, 

 yet giving the observer a chance to prevent many accidents 



* The error avoided by its use is probably no greater than this, but it 

 seemed best to substitute a measureable error for a more uncertain one. 



f This method was adopted from J. A. Harker, Specific Heat of Iron at 

 High Temperatures, Phil. Mag., x, 430, 1905. 



