720 
peculiarity lay in the nature of its radiation. 
To test this a prism was placed before the 
lens of the camera which broke up the 
image of the spark into a series of spark 
images of different color. The plate was 
exposed to the spectrum flash of a single 
spark, then removed from the camera and 
exposed to the candle light, and developed. 
If the reversing effect was due to any pecu- 
liar radiation or wave length, we should find 
the reversal at the part of the spectrum 
where the effective radiation belonged, say 
in the infra red, if the reversing power lay 
in long waves given out by the spark. It 
was found that the entire spectrum came 
out lighter on the negative than the fogged 
background. A second plate was exposed 
to the spectrum flash, then slightly fogged, 
and a second spectrum impressed on it. On 
developing one spectrum came out light and 
the other dark, as shown in Fig. 2. Clearly 
the effect does not depend on wave length. 
It then occurred to me that the time element 
might enter into the problem. The light 
of the spark is over in about 1/50000 of a 
second and it did not seem impossible that 
a bright light of exceeding short duration 
might act quite differently on a plate from 
a weaker light of longer duration. This 
may be tested in a variety of ways. We 
may open the lens wide, impress the image 
of a single spark on the plate, and then stop 
the lens down and superimpose a number 
of spark images sufficient to make the total 
exposure the same in each case. This was 
the first method which I tried. In order 
to compel the successive sparks to pass over 
the same path, that their images might be 
superposed, I shut them up ina capillary 
tube. With the lens open wide enough to 
give the maximum reversing action, I passed 
a single discharge through the capillary. 
Stopping the lens down to one-quarter of 
its former aperture, four discharges were 
passed through the tube. The plate was 
then fogged in the usual manner, and on 
SCIENCE, 
[N. 8. Von. X. No. 255. 
development the single discharge was re- 
versed, but the composite one was not. 
Fig. 3 is from a plate showing this effect. 
The upper images are those of single dis- 
charges through the capillary, with differ- 
ent apertures on the lens ; the lower images 
are those of double or triple discharges 
through the same tube. The left hand side 
of the plate was exposed to the candle light 
for different amounts of time, by moving 
the screen over small distances during the 
exposure. Only the single discharges re- 
verse, though the density of the images on 
the unfogged portion of the plate is the 
same. This is very strong evidence that 
the duration of the illumination was the im- 
portant factor. 
Some years ago I measured the duration 
of the flash of the exploding oxy-hydrogen, 
finding if about 1/12000 of a second. 
Possibly the flash of such an explosion 
would duplicate the effect. I exploded 
several glass bulbs filled with electrolytic 
gas, but found that the action was the same 
as that of ordinary light, it being impossible 
to get any reversal. The flash evidently 
lasted too long, or there still remained some 
undiscovered factor. 
The difference between the action of 
spark light and the light of the oxy-hydro- 
gen flash is shown in Fig. 4. Plate ‘a’ 
shows the effect of the explosion flash. 
Squares 1 and 2 received the light from an 
exploding bulb, the rest of the plate being 
covered. Squares 1 and 3 were then ex- 
posed to the light of the candle. Square 1, 
which has received the light from both 
sources, is the brightest, that is, the effects 
_ are additive, there being noreversal. Plate 
‘b’ shows the action of the light from the 
spark. Squares 1 and 2 were illuminated 
by the spark light, then squares 2 and 4 
were exposed to the candle. In this case 
square 4, which was illuminated by the can- 
dle, is brighter than square 2, which re- 
ceived both the spark light and candle 
