104 REPORTS ON INVESTIGATIONS AND PROJECTS. 



3. A light swinging shield under the furnace controls the exposure of the 

 calorimeter. It performs with certainty, ease, and quickness the functions 

 often sought through a movable calorimeter or furnace. 



4. Temperatures uniform and constant to 0.1 at least are obtained for the 

 interval from 200 to 600 in an electrically heated and stirred bath of melted 

 metal, which surrounds but does not touch the body to be heated. 



5. A steam-heated oven is also used. With this a simple automatic ar- 

 rangement gives uniformity of results in dropping bodies into the calorimeter. 



6. This paper makes a short addition to a previous discussion of equi- 

 potential leakage shields. By partly inclosing the apparatus in cases which 

 also contain drying material, and then coating with paraffine all exposed in- 

 sulating surfaces, insulation was assured, without inconvenience, under 

 atmospheric conditions previously considered almost fatal to high accuracy 

 in the measuring system. 



(24) A test of calorimetric accuracy. Walter P. White. Phys. Rev. Dec. 1910. (In 



press.) 



The two calorimeters described in the previous paper were calibrated by 

 the electrical method, and the determinations were varied considerably, so as 

 to afford information as to the errors present. 



An automatic switch which started and stopped the heating current in 

 obedience to electric signals from a chronometer was an advantageous sub- 

 stitute for a chronograph ; otherwise, the details of the calibration were much 

 as usual. The results were : 



1. The average accidental error of a complete calorimetric determination 

 has been reduced to about one part in 30,000 ; the systematic variations under 

 rather diverse conditions were no greater than 0.0002. The essential con- 

 ditions for this performance appear to have been : ( I ) ordinary care in 

 manipulation and observing, (2) an electrical thermometer reading consist- 

 ently to 0.0003 , (3) auxiliary electrical apparatus accurate to 0.00003, (4) 

 a complete and uniform inclosure of the calorimeter, (5) regulation of the 

 stirring, (6) elimination of evaporation, and (7) the use of unusually large 

 temperature intervals. Of these features, only the last two are specially 

 characteristic. 



2. The average accidental temperature error was brought below 0.00025 °. 

 This is practically a perfect performance for the multiple thermo-element 

 used as a thermometer. 



3. No error due to the cooling correction could be detected, even where the 

 total average error was only 0.00003. 



4. The preponderant calorimetric error was in the temperature measure- 

 ment. 



5. The use of a large temperature interval obviously diminishes the effect 

 of the temperature error. It brings no corresponding disadvantage. Inter- 

 vals much larger than those ordinarily used are therefore to be recommended 

 where accuracy is desired. 



(25) The sulphides of iron. E. T. Allen, John Johnston, and J. L. Crenshaw. Crystal- 



lographic study by E. S. Larsen. Am. Journ. Sci. (In press.) 



i. Pyrrhotite has been prepared in the massive form by the decomposition 

 of pyrite or by heating sulphur and iron together to a temperature of about 

 700 . The dissociation of pyrite into sulphur and pyrrhotite begins very 

 slowly about 500 , and the vapor pressure (judging by the strong heat ab- 

 sorption) reaches one atmosphere at about 68o°. The change is reversible 



