716 



Dr. S. J. Allen on a Null Instrument 



the instrument. This curve is shown in fig. 2, A, the 

 abscissae representing the position of the pointer on the 

 scale s in millimetres, and the ordinates the rate o£ leak 















Fig. 



2. 

































/ 







90 



1 

















/ 









4 















/ 



/ 



7^ 









1 













V 



j 



& 











55 













/ 



/ 













^ 

 3 











/ 



[/ 













70 



I 











/ 



r 























/ 



/ 























i 



















50 









/ 



i 























// 

























/ 



/ 























// 



























// 

























I 



/ 

















I 







/ 



' 























/ 



























/ 























10 



l 



I 



7 























J 



/ 















-Sea 



7a7('dfzd/n^- 



10 2o 30 40 SO 



Jo too no izo 



calculated in per cent, of the maximum. That is, the rate 

 of leak when the cover is opened to its full extent is taken 

 as 100 per cent., and the others calculated accordingly. 



The mean error of this calibration curve is probably not 

 greater than one or two per cent. 



It was found that a balance-point could be readily esti- 

 mated to about one-half a millimetre, and with extreme care 

 to one-tenth of a millimetre, the error of this being much less 

 than the probable error of calibration. 



For more accurate work other methods of calibration could 

 be devised which would give a more exact calibration curve. 

 One method which would probably do very well would be to 



