AND STATICS UNDER THE INFLUENCE OF LIGHT. 
353 
chain of the generator, but a modification reduced the variations to about Ik per cent. 
The axis x (fig. 6) on which the ball L turns, was filed up till the movement of the ball 
in the axis was easy, and (3, which guides the balls, as 
well as the stem of L at a, were then filed and clamped 
to give the ball a perfectly free movement up and down, 
but not much side play. Sheets of lead (s) were then 
fixed round the stem as near to the balls as possible, and 
the weight carefully adjusted so that a small additional weight of about 10-15 grammes 
placed near the balls should draw them quite down and open the valve, and that the 
removal of this weight should bring the balls up again, closing the valve. The 
modified generator gives, without a balance governor, for the heights of the bell out 
of the tank, between 16'5 centims. and 5 centims. (these are the limits within which 
the bell of the gas-holder chiefly varies in its height during the production and con¬ 
sumption of gas), a variation of only l - 5 millim. for 100 millims., i.e., of 1^ per cent. 
These variations of the pressure were further reduced by the balance governor. 
It was found that with a burner of such a candle power as I had to use (about 
12 x 20 = 240 candles), a pressure of about 4 inches is necessary (instead of the 
usual 2). For this the upper chamber of the gas-holder had to be almost filled with 
water, leaving only space sufficient for the expansion of the water (in the winter care 
must be taken that the water does not freeze when the gas-holder is in the open air ; 
more free space or a salt solution must be used). The gas generator was placed on the 
roof of the laboratory; from the generator a lead pipe brought the gas to the room ; 
first to the balance governor, from the balance governor to the regulating tap, from the 
regulating tap to the water manometer WM, and then to the purifier ; here it passed 
over the purifying substance of the lower cylinder, then of the higher cylinder, and 
thence to the burner. 
The Regulating Tap to counteract the different sources of error was of the ordinary 
type. One tap, «, must always be quite open, when acetylene is consumed during 
the experiments. By turning another tap, the quantity of gas passing per unit of 
time to the burner is regulated, and the adjustment is indicated by the pointer on a 
fixed scale of 90 degrees. The purpose of this tap was to adjust the supply of gas 
to the burner so that the intensity of the acetylene light should, under varying con¬ 
ditions, always be kept the same. A series of conditions affect the intensity of the 
acetylene flame and necessitate the use of a regulating tap, such as the variation in 
the diameter of the outflow tubes in the nipples of the burner or the variation of the 
atmospheric pressure, which though in one and the same day hardly ever varying more 
than I or 2 millims., changes during a longer period considerably. 'There is another 
source of error of an irregular but temporary nature which also necessitates its use : 
the admixture of air when a new container is used. Evervthing was done to remove 
this source of error at its source. When a new container was introduced, the tap 
supplying water to the same was opened, and the cross-bar of the container pressing 
VOL. CXCIX.--A. 2 Z 
Fig. 6. 
