EDINBURGH MEETING. 
515 
the group. Thus J and E are united by two bonds, while one would be sufficient. 
We have an instance of this kind of addition in the union of chlorine and ole¬ 
fiant gas. The two carbon atoms of olefiant gas are doubly-related to one 
another; by the action of chlorine one of these relations is broken, and each 
carbon atom becomes related to a chlorine atom. 
Addition may take place in another way, namely, by an increase in the number 
of relations of an atom; thus in ammonia, nitrogen is trebly-related, being united 
to three atoms of hydrogen, but when ammonia is brought into contact with hy¬ 
drochloric acid, chloride of ammonium is formed, the nitrogen becoming fivefold- 
related, being united to four atoms of hydrogen and one of chlorine. We may 
illustrate this by supposing that John, who is already doubly-related, borrows 
£1000 from Robinson, and buys another house, thus entering into two new rela¬ 
tions, one to his creditor and one to his investment. Now if we offer Robinson a 
better investment, he will call up his money from John, who must either borrow 
from some one else or sell one of his houses. Both cases occur in chemistry ; if we 
mix caustic potash with chloride of ammonium, we offer, so to speak, the better 
investment of potassium to the chlorine, which accordingly leaves the nitrogen, 
but the nitrogen, at the same time, loses one atom of hydrogen, just as John 
parted with his creditor and one of his houses together ; but if we try a similar 
experiment with the chloride of a complex ammonium (such as chloride of 
tetramethylammonium), mixing it, say, with moist oxide of silver, and thus 
tempting the chlorine to leave the nitrogen and unite with silver, the nitrogen 
does not lose a methyl atom along with the chlorine, but, if I may use the ex¬ 
pression, pays off the chlorine by borrowing from oxygen. This analogy might 
be carried much further, but it is unnecessary to do so. My object is attained 
if I have shown that any kind of relation which can be single, double, treble, 
etc., may be used to illustrate chemical constitution, and that we are not neces¬ 
sarily restricted to geometrical relations. 
Our time does not allow me to dwell longer on this part of the subject; I 
shall, therefore, proceed to the consideration of the relation between chemical 
constitution and physiological action. 
The difficulty of determining this relation depends mainly on our ignorance 
of the constitution and action of the great majority of substances, which makes 
it impossible for us to make much progress by direct comparison of constitution 
with action. All that has been made out in this way is, that the soluble com¬ 
pounds of what we may call a poisonous element, such as lead, have all nearly 
the same action, and similarly of a poisonous radical, such as cyanogen. Here, 
however, we meet some remarkable exceptions; thus, kakodylic acid, although 
readily soluble, and containing a large amount of arsenic, is inert, and a number 
of the double cyanides, such as ferrocyanide of potassium, show no trace of the 
action of hydrocyanic acid. 
In this difficulty it occurred to Dr. Fraser and myself, that another method 
of inquiry might lead to better results. The method which we have adopted 
consists, first, in introducing a known change into the constitution of an active 
substance ; and, second, in comparing the action of the new product so formed 
with that of the original substance. 
The first set of substances to which we applied this method (in an investiga¬ 
tion communicated to the Royal Society of Edinburgh, January 6, 1868) was 
the group of vegetable alkaloids, and we have treated in this way strychnia, 
brucia, thebaia, morphia, codeia, and nicotia. Each of these alkaloids contains 
a trebly-related atom of nitrogen, which can easily be made fivefold-related. 
We do not know what its original three relations are, but we know what the 
two new ones are which we introduce. Thus, confining our attention to strych¬ 
nia (and what is said of it applies with slight modification to the others), we 
find that a molecule of that alkaloid contains two atoms of nitrogen; of the con- 
2 L 2 
