MR. W. CROOKES OK REPULSION RESULTING PROM RADIATION. 
271 
at the time in my note book, and it is my custom to verify everything of importance. 
The record of experiments given in paragraph 245 was written in May last. The 
repetition now described took place in October. 
When a lighted candle was placed 1^ inches from the bulb continuous rotation was 
produced, but instead of the chromic oxide being repelled the selenium was repelled. 
Another candle was lighted, and the two only made the abnormal rotation stronger. 
After a time the bulb became warm and the rotation got slower till it stopped. The 
candles were blown out, and when the bulb had cooled,, a water cell was interposed 
and one candle was lighted. Good rotation now took place in the same direction 
as when no water intervened ; the radiometer, in fact, behaved just like an ordi- 
nary lampblack and white pith one, the scarlet selenium representing the black 
surface. The results were not quite so strong as when lampblack is used, but 
they were so definite that had it not been for the previous observations I should 
unhesitatingly have assumed that the true action was as now recorded. But knowing 
that my former results were just as carefully worked out as the present ones, and 
not admitting “ experimental errors ” as explanations of anomalies, I sought long and 
anxiously for a physical cause of these two diametrically opposite results, Every 
circumstance which could influence the result was the same in the two experiments, 
yet in May I found a candle repelled chromic oxide more than it did selenium, whilst 
in October I found the selenium decidedly the more repelled of the two, positive 
rotation occurring in one case and negative in the other. 
256. Next day I repeated the experiment by placing a lighted candle near the 
radiometer, and the fly at once rotated, the chromic oxide being repelled. Nothing had 
been moved, nothing had changed since the previous day except the candle. Here 
was a clue which, followed up, explained everything. The candle used in May was a 
large wax candle, four to the pound ; that used in October was a sperm candle of 
a smaller size, six to the pound, and having a whiter flame. I found I could 
always get the chromic oxide repelled by using the wax candle, and the selenium 
repelled by using the sperm candle ; the difference between the two is evidently 
due to a variation in the proportion of light to heat. The sperm has a very white 
flame, whilst the wax candle burns with a yellower flame : for equal amounts of 
light, therefore, the wax candle gives more heat ; and according to theory heat is 
principally required to repel chromic oxide, whilst light chiefly repels selenium. To 
test this by experiment was easy. A spirit flame was brought near the bulb ; 
decided repulsion of the chromic oxide with quick rotation followed. A Bunsen 
burner, with the air holes covered (giving a luminous flame), brought near the bulb 
caused repulsion of the selenium, but scarcely strong enough to give continuous 
rotation. On uncovering the air holes and making the flame non-luminous the 
chromic oxide was strongly repelled, giving rapid rotation. 
The radiometer was exposed to the light from two candles behind a transparent 
plate of alum, 5 milliins. thick ; continuous quick rotation ensued, the selenium being 
