600 
MESSRS. W. SPOTTISWOODE AND J. FLETCHER MOULTON 
spark is small and therefore the intermittence very rapid. It is in such cases difficult 
to decide satisfactorily whether the bright portions are absolutely continuous or not. 
But this difficulty is in reality of very slight importance, inasmuch as there is no doubt 
that the other portions of the phosphorescence are intermittent, and these brighter 
spots do not differ from the rest of the phosphorescence in the origin of their lumi¬ 
nosity, but only in their being more favourably situated for the concentration of 
molecular streams upon them. Hence there can be no reasonable doubt that this 
apparent continuousness is due to the persistence of the phosphorescence. In the 
case of large air-sparks the mirror when revolving at a high velocity exhibits a slight 
haziness on one side of the image of the slit, showing that the dying out of the phos¬ 
phorescence is not quite instanstaneous, though very nearly so. 
It will thus be seen that the phosphorescence produced by the separate discharges 
of an intermittent current must be very intense, seeing that the periods during which 
the glass is brightly illuminated by it are extremely short compared with the periods 
that intervene, and yet the intensity of the apparent illumination of the glass is 
greater than with the continuous current. This shows that the particles must be 
driven off at a greater velocity or in greater numbers during the short period occupied 
by an individual pulse of the intermittent discharge than during the continuous dis¬ 
charge—a difference that would naturally follow from the fact that the electricity in 
each of these individual discharges represents the total accumulation of the period 
between two discharges. But it is remarkable that the vanes of a radiometer, when 
used as the negative terminal of a vacuum tube, as in Crookes’ experiments, revolve 
very much more quickly under the influence of a continuous discharge than an inter¬ 
mittent one. If an air-spark of increasing length be introduced into the current of 
a Holtz machine that is driving a radiometer electrically, the driving power wall 
gradually diminish, and ultimately cease. This is, perhaps, equivalent to saying that 
the action upon the radiometer is dependent on the quantity of the discharge, and not 
upon its tension, since, of course, the introduction of a long air-spark into the circuit 
must necessarily have the effect of enormously decreasing the quantity of electricity 
passing, although it similarly increases the tension. But then we are met with the 
difficulty that the intensity of the phosphorescent illumination in similar cases is 
increased by the introduction of an air-spark in spite of the diminution of the quantity 
of the discharge. A possible explanation of this law is obtained by supposing that 
the effect on the radiometer is proportional to the 7nomentum with which the particles 
leave the terminal while the phosphorescence depends on their energy. This would 
mean that a diminution of quantity diminishes the number of particles leaving the 
terminal while the increase of tension increases the velocity; such increase being 
insufficient to prevent the total momentum of the particles from being decreased by 
the introduction of an air-spark, but sufficing to cause a marked increase in their total 
energy. 
It is a very significant fact that the intermittence of the phosphorescence exists 
