1913] KNIGHT & CROCKER—TOXICITY OF SMOKE 359 
methane, and carbon monoxide, and in tobacco smoke pyridine, 
ammonia, hydrocyanic acid, and nicotine in addition, leaving 
entirely out of consideration the tars which have a relatively low 
toxicity where only their vapors are involved, and probably no such 
. a magnitude of toxicity under any conditions as several of the 
carbon-bearing gases. Of these substances we will give detailed 
experiments only on carbon monoxide, hydrocyanic acid, and 
nicotine. For the other substances it will suffice merely to cite a 
portion of a table to appear in one of our later papers, along with 
the details from which it is derived. We consider the details on 
carbon monoxide here because it is the more abundant toxic gas 
and because Moriscu suggests that it may be the one rendering 
the smoke so toxic to seedlings. 
Experiment VI.—Effect of carbon monoxide 
It is first desirable to make sure that the carbon monoxide used 
is free from the noxious gases, or at least that the effect produced 
is due to the contained CO and not to some impurity. For this 
reason the carbon monoxide was generated by three different 
methods and the three products compared as to their effects. It 
is assumed that if equal amounts of the three sorts of gas produce 
equal effects, the effect is due to the CO and not to impurities. Since 
the heavy hydrocarbons are so toxic, it was thought well to see 
whether washing the CO in bromine would reduce the toxicity. 
When oxalic acid was heated with several times its weight of 
concentrated H,SO, and washed with 40 per cent NaOH, a gas 
resulted which gave (duplicate analysis) 99 per cent absorption 
with ammoniacal cuprous chloride. Potassium ferrocyanide was 
heated with 8-10 times its weight of concentrated H,SO, and washed 
with 40 per cent NaOH. In duplicate this showed 96 per cent CO. 
Sodium formate was heated with concentrated H,SO,and produced a 
gas giving 89 per cent absorption with ammoniacal cuprous chloride. 
The epicotyls were 2.5-3.5 cm. tall at the beginning of the 
experiment and were inclosed in 1o-liter cans. The following data 
show the sources and concentrations (correcting for impurities) 
of CO used, and the condition of the seedlings at the close of the 
experiment. 
