1917] KRAYBILL—ALKALI SALTS 51 
alkali carbonates, it would be difficult to explain by NEssLEer’s 
theory why lithium and sodium carbonates are not effective, and 
caesium, rubidium, and potassium carbonates are effective. 
GARNER (4) suggests that the carbonate might act favorably 
by alternately giving up and taking on carbon dioxide. We have 
seen that caesium, rubidium, and potassium carbonates are very 
much more effective than the carbonates of sodium and lithium. 
At joo C. the order of the alkali carbonates according to their 
increasing ease of dissociation is as follows: potassium, rubidium, 
sodium, caesium, and lithium. Lithium carbonate is dissociated 
into carbon dioxide and lithium oxide to a much greater degree 
at 7oo° C. than any of the other carbonates (9). If potassium 
carbonate acts favorably by alternately giving off and taking up 
carbon dioxide, it is difficult to see why lithium carbonate should 
not be even more effective. It seems, therefore, that this theory 
will not explain the beneficial action of the caesium, potassium, and 
rubidium carbonates. 
From table VIII we see that all of the salts which were effective 
in increasing the fire-holding capacity of tobacco will cause the 
lump of sugar to burn with a flame when ignited by means of a 
gas burner. Some of the salts which were only slightly effective 
in increasing the fire-holding aia d of + AED, such as nthe 
the combustion of sugar with the production. of a flame. ‘Here 
again caesium, potassium, and rubidium cecuutes are particu- — 
larly active. Barra (1) has suggested that the salts present in — 
the leaf may aid in the combustion by raising the temperature 
of the leaf, the effect being somewhat analogous to the effectof 
salts upon raising the boiling point of water. As shown i in table : 
VIII, such substances as metallic fili | ind 
 bonized sugar when used in larger pial also ar a. 
: eee et ee to burn. From these eaten 
> sugar, it would seem as though their effect in raising — 
