spark between two rounded ends of platinum wire Y2 mm. diameter sepa- 

 rated a small fraction of a mm. The potential at the spark gap was not 



PL/^T£. a. 



PL RTE. X. 



great enou.uli to spark a distance of % mm. While working with this 

 apparatus a phenomenon occurred which shows how easily constant errors 

 may influence the result. The lead screen used to intercept the radia- 

 tion was suspended by cords to the top of the iron lamp chimney so as 



