138 Mr. Dyke on the Determination of the Mean Spherical 



pings, supplying current to motor and lamp respectively, 

 and also a pair for measuring the voltage on the lamp- 

 terminals. All the electrical measurements were accurately 

 made on a potentiometer. 



The slip-rings are connected by flexible leads passed 

 through the hollow shaft to the terminals of the lamp- 

 holder. A cord grip P relieves the motor and brush sup- 

 ports from the drag of the flexible cord. 



The carriage is pulled along the photometric bench by an 

 endless cord passing over pulleys at the ends of the b^nch 

 and gripped by a cord grip C. The position of the lamp 

 on the bench is read by means of the wire Q. 



The advantage of this rotator is that it is self-contained , 

 and hence can be arranged so as to enable a fixed photo- 

 meter-screen to be used, making it much more convenient 

 to work with. 



The method employed was to fix the lamp in the holder 

 with the filament so placed that the axis of the block on 

 which the filament was formed coincided with the axis of the 

 photometric bench. 



This was called the zero position. 



The voltage on the lamp terminals was kept at that re- 

 quired to give the marked candle-power in the zero position. 

 Readings of candle-power were then taken at frequent 

 intervals round the lamp, special attention being paid to 

 those positions in which the candle-power was changing 

 rapidly. 



Finally the lamp was rotated, and a series of readings of 

 the mean horizontal candle-power taken. 



The value thus obtained was checked against that given as 

 the mean ordinate of the curve plotted from the step-by-step 

 readings; the results in almost every case agreeing to within 

 three-quarters of 1 per cent. 



To obtain rapidly the relation between mean spherical 

 and mean horizontal candle-power, a modification of the 

 method devised by Mr. C. P. Matthews was employed. 



This method consists in placing several pairs of mirrors 

 round the circumference of a semicircle, and so arranging 

 them that light emitted from a source placed at the centre of 

 the system shall be incident on a white screen, also situated 

 at the centre, at the same angle with the vertical as that which 

 it made on emission from the source. Then, if the source is 

 symmetrical about the diameter of the semicircle, and if 

 Lambert's cosine law holds for the screen used, it can he 

 shown that the intensity of illumination on the screen is 

 proportional to the mean spherical candle-power. 



