402 BOTANICAL GAZETTE [DECEMBER 
concluded that both the molecules and the ions, both negative 
and positive, are devoid of toxic action. 
The two next columns of this table show the ions into which 
the salts are separated. Here then is proof that to the fungi 
under investigation the cathions Mg, Ba, Na, and K are non- 
toxic. The anions SO,, Cl, Br, I, MnO,, and C,H,O, are non- 
toxic at the strength used. The second half of table XXI 
shows those salts which were found to have toxic action. 
As Gl-ions have previously been proven non-toxic the effect 
must rest with the molecule of HgCl, or with the ion of Hg*. 
The salt at the strength used was almost completely dissociated, 
hence the effect is due to the Hg* ion, which was the most pow- 
erful one experimented with. 
With the two acids, HCl and H,SO,, the anions had proven 
non-toxic; so, by similar reasoning, the poisonous property rests 
with the H+ ion. As H,SO, contains twice as much H* as 
does HCl it should have twice its toxic power in equi-molecular 
solutions. This statement has met no adverse results in experi- 
ment, but there has been slight, though by no means positive 
evidence, to sustain it. 
With KCN the cathion is non-toxic. The CN- then must be 
poisonous, and it has about the same toxic action as does mn 
With all the copper salts the anions are non-toxic, hence the 
poison is in the molecule or in the copper, and in the more dilute 
solutions it undeniably rests with the copper. As each copper 
compound has as many atoms of copper per molecule as the others 
it would be equally toxic. This expectation is almost fully met, 
with the exception of copper acetate for Penicillium and Uromy- 
ces, and of copper sulfate for Macrosporium. These three slight 
deviations from the theory stand against thirteen observations 
with the copper salts tending to support it. j 
In the potassium chromate and bichromate solutions the pol 
sonous action rests evidently with the anions. The bichromate 
is in every case more poisonous and in most cases far more 
poisonous than the chromate, as may be seen by tables XII and 
III. 
' 
