28 THE AUSTRALIAN NATURALIST. 
(b) A colloid belonging to the suspensoid class gives, 
with water, a mixture which is non-viscous and 
non-gelatinising, but is coagulated on the addi- 
tion of small quantities of electrolytes. <A col- 
loid belonging to the emulsoid class gives, with 
water, a mixture which is viscous, gelatinises, 
and is not readily coagulated by salts. 
With a colloid belonging to the former class, an 
amount of salt, which is not capable of causing immediate 
coagulation, is, nevertheless, effective after a certain inter- 
val, and the total quantity of electrolyte required to bring 
about complete coagulation of the suspensoid varies ac- 
cording as the electrolyte is added, all at once or jn several 
portions successively. 
(¢) The Danysz Phenomenon.—tThe toxicity of a mix- 
ture of diphtheria toxin and antitoxin depends 
on the way in which the two are mixed. Tf the 
amount of toxin added is such that the mixture 
is untoxic, then in a second experiment, in which 
the same amounts of toxin and antitoxin are 
taken, in which, however, the toxin is added in 
instalments, the resulting mixture is toxic. ‘This 
phenomenon is exactly analogous to what hap- 
pens in the precipitation of a colloid by an elec- 
trolyte, or in the precipitation of one colloid by 
another; the amount of electrolyte or colloid re- 
quired for complete precipitation varies axecord- 
ing as it is added all at once or in instalments. 
The condition of a toxin in antitoxin mixture, 
therefore, resembles that of colloidal solutions in 
that it is not completely defined by a statement 
of its composition; its character depends on its 
previous history. 
Now, in the case of a crystalloid, its character is con- 
stant, for example, a certain amount of acid neutralises a 
certain amount of alkali, and it does not matter whether 
the acid is added all at once or little by little, the same 
amount of acid is required in each case to effect neutrali- 
sation, and the products in each case are identical. 
But in the case of a colloid, since its character de- 
pends on its previous history, and since each event in its 
previous history may have influenced its character, I see 
no difficulty in conceiving that contiguous colloids (in ani- 
mal cells) influenced by different external stimuli to act 
reflexly may, in time, influence each other, and so act 
