3718 
it into the large air tank, where it was stored for use. By this means 
the supply of gas for a complete test could be stored up before the test 
itself was started. One more precaution had to be taken in using this gas, 
as it could not be allowed to affect the temperature of the coal as it 
passed through. To prevent this, a coil of pipe was placed over a gas 
flame and the gas passed through and heated up to the temperature of 
the coal, before it was allowed to enter. The temperature of the gas was 
measured by a thermo couple which extended into it through a tee in the 
pipe line. To make assurance doubly sure the end of the cartridge itself 
was filled with steel chips, as is shown at 22 in the sketch in Fig. I. The 
entering gas was thus forced to pass through a considerable volume of 
these chips before coming into contact with the coal. It was found difficult 
to heat the gas up to the highest temperature of the producer. This 
might tend to affect the seeming temperature at which the last traces of 
tar appear. The tendency of the gas would always be to be lower than 
that of the coal. For this reason the end of the thermocouple was placed 
in the coal at the end where the gas enters it, and therefore at its cooler 
end, in case there should be any difference at all. Thus the temperature 
reported as the one at which the last trace of tar appears is as accurate 
as it is possible to make it. 
5. To test the gases for tar. The next problem was to find a means 
of subjecting the gases from the coal to a continuous test for tar. The 
most searching and satisfactory test known to the author, and the one 
used by the gas companies over the country, is to allow a small stream 
of the gas to strike a piece of white paper at a high velocity. If there 
is any trace of the tar at all in the gas, it soon leaves a spot on the paper. 
This test was adopted. ‘To use it, the gas must be cooled down before it 
strikes the paper. This was accomplished by keeping a cloth filled with 
cold water constantly lying on pipes (12) and (18). It was desired to 
have the test continuous. The device used to accomplish this is shown 
in Figs. 1 and 2. The rollers (18) are about 10” in diameter and are sup- 
ported by steel rods through their centers which turn freely in iron sup- 
ports at either side. On one end of each of these steel rods was placed a 
small wooden spool, around which was wound a cord, supporting weights W:; 
or W.. A long strip of cloth was wound around one of these rollers and 
its end started around the other. A piece of paper ribbon was wound on 
with the cloth. W: and W, tend to turn the rollers in opposite directions, 
thus keeping the cloth and paper strip tight. W: is enough heavier than 
