DUNSTAN, RICE, KRAUS: SPECTRAL LINES OF SODIUM. 81 
gradually increased the visibility passes through successive maxima 
and minima. These maxima will touch the curve given by 
be 
v—A2 4” and hence steadily decrease as the difference of path 
becomes greater. Hence this envelope, determined by observation 
at points of maximum visibility will give a curve expressed by 
D2 
A2 4” and from this curve may be determined A and /\; where 
/\ is the abscissa of the point at which V=™%A and A is the visi- 
bility at the point where Do. 
It may be proved further that /\ is connected with 6, the half 
loge 2 
width of the line in the source by the expression 6 and 
mL\ 
hence from the value of A may be determined the width of the 
line. 
The magnet used was a large one with movable pole pieces. It 
was capable of carrying a current of fifty amperes for short periods, 
and this current was used in some of our experiments. The space 
between the pole pieces varied from 3 to .75 cm. and thus gave a 
great range of field intensities. Field intensities were measured by 
means of exploring coils of known area, which were suddenly 
jerked out from between the pole pieces. The throw of a Hori- 
zontal Pattern D’Arsonval Ballistic Galvanometer, which was con- 
nected through a resistance box to the exploring coil, gave data for 
the calculation of the strength of the magnetic field. In order to 
avoid error due to damping, which in this form of galvanometer 
is considerable, the exploring coil was arranged so that it broke the 
circuit immediately upon being jerked out of the field and hence 
left the damping of the galvanometer the same as upon open cir- 
cuit. By changing the resistance in the box the galvanometer 
throws were kept nearly constant in magnitude and hence the errors 
due to damping, reduction to arc, etc,, were avoided. 
The galvanometer constant was determined by means of an 
Elliott standard condenser and two Carhart-Clark cells. 
For each position of the pole pieces a curve, whose co-ordinates 
Were strength of field and exciting current, was drawn, thus enab- 
ling the strength of field to be interpolated from an ammeter read- 
ing. The two exploring coils used were wound with a single layer 
of No. 36 wire and had total areas of 102.37 and 34.99 square cen- 
timeters respectively. 
\4 
1] 
1 
