482 



SCIENCE. 



[N. S. Vol. VII. No. 171. 



that the vacuum, or ether line, locates itself 

 exactly the same in all. 



In making the above described observa- 

 tions, I looked for some change in the phe- 

 nomena when the exhaustion reached the 

 point at which the mean free path of the 

 gas molecules equalled the distance between 

 the thermometer bulb and the cold walls of 

 the enclosing globe. This should have 

 been at a pressure of about two millionths. 

 ]S"o such change was observable, however, 

 in any case. Partly in pursuance of the 

 same idea, I resolved to repeat some of my 

 experiments, using a very much smaller 

 radiation bulb. This I expected would also 

 reduce that portion of the total cooling 

 effect due to convection currents. I accord- 

 ingly employed the bulb of tube P, Fig. 1, 

 in my further experiments. This is made 

 from a thin glass tube slightly less than 20 

 millimeters internal diameter, and in it hangs 

 the same thermometer A which was used 

 before. In transferring the thermometer 

 great care was taken to avoid any dis- 

 turbance of the coating of lampblack on its 

 bulb. At b is a contraction of the tube P, 

 to prevent the thermometer bulb swinging 

 against the inside of the tube. The con- 

 traction b is, however, much larger than 

 the thermometer stem, so that normally the 

 latter does not touch it. The thermometer 

 bulb hangs exactly in the center of P, near 

 its bottom, and is separated from it by a 

 space of a trifle more than six millime- 

 ters — -almost exactly a quarter of an inch— ■ 

 instead of two inches, as in the case of the 

 ' Large bulb.' The tube or bulb P, I shall 

 hereafter designate the ' Small radiation 

 bulb,' or simply ' Small bulb,' to distinguish 

 it from the large one. 



The curve for hydrogen, with the small 

 bulb, dififers radically in size and form from 

 that obtained with the large bulb. Section 

 A, instead of drooping rapidly with decreas- 

 ing pressure, maintains almost full value 

 throughout. Section B starts with nearly 



double its old value, but breaks down much 

 earlier. Section C starts with a little higher 

 value, but is much straighter, and conse- 

 quently has a lower value throughout most 

 of its length. The curves a, aa, aaa are very 

 peculiar. They start at atmospheric pres- 

 sure with much smaller total and very, dif- 

 ferent relative ratios than before, and are 

 successfully absorbed into A. They reap- 

 pear later, however, but with small ratios. 



Fig. 3 gives the curve for air, with the 

 small bulb. It differs from that with the 

 large bulb quite as much as did the hydro- 

 gen curve. Section A droops slightly, and 

 regains almost its full atmospheric value 

 at one per cent, pressure. Section B has 

 the same form as with the large bulb (Fig. 

 2), but more than double its value; and 

 section C also has a much higher value 

 throughout. The curves a, aa, aaa have 

 small ratio values at thebeginning, and are 

 absorbed into section A, the same as with 

 hydrogen. But aa and aaa coalesce when 

 they reappear, and coincide to the end ; 

 while the ratio between a and aa remains 

 constant at a very small value. 



The curve for carbon dioxide, with the 

 small bulb, closely resembles the air curve 

 in form, but has a very much smaller value 

 throughout. While the curves aa and aaa 

 are soon united, and remain so to the end, 

 a and aa never disappear as they did in the 

 cases of hydrogen and air. 



With the small bulb, as with the large, 

 no change in the character of the phenom- 

 ena was observable when the exhaustion 

 had reached the point at which the mean 

 free path of the molecules equaled the space 

 through which the heat was conducted. 

 This point was reached in the small bulb at 

 a pressure of about fourteen millionths. 



It seems reasonable to assume that the 

 radical difference between sections A of the 

 curves obtained with the large and small 

 bulbs respectively was due to an almost 

 complete suppression of convection currents 



