326
ON THE THEORY OF GROUPS.
[690
where, for shortness, ab, ace, &c., are written instead of (ah), (ace), &c., to denote the
cyclical substitutions a into b, b into a; and a into c, c into e, e into a, &c.; and
it is at once seen that by the same substitutions the lines may be derived from any
other line.
It will be noticed that in the foregoing substitution-group each substitution is
regular, that is, composed of cyclical substitutions each of the same number of letters;
and it is easy to see that this property is a general one; each substitution of the
substitution-group must be regular.
By what precedes, the group of any order composed of the functional symbols is
replaced by a substitution-group upon a set of letters the number of which is equal
to the order of the group, and wherein all the substitutions are regular.
The general theory being thus explained, I endeavour to form a substitution-
group with the twelve letters abcdefghijkl; and I assume that there is one substitution,
such as abc.clef.ghi.jkl, and another substitution, such as agj. bfi. cek. dhl. Observe
that, if the twelve letters are to be thus arranged in two different ways as a set
of four triads, without repetition of any duad, all the ways in which this can be
done are essentially similar, and there is no loss of generality in taking the two sets
of triads to be those just written down. But the substitution to be formed with either
set of triads will be different according as any triad thereof, for instance agj, is written
in this form or in the reversed form ajg. There are thus in all sixteen substitutions
which can be formed with the first set of triads, and sixteen substitutions which can
be formed with the second set of triads; and the relation of a triad of the first set
to a triad of the second set is by no means independent of the selection of the
triads out of the two sets respectively. To show this, take the two substitutions quite
at random; suppose they are those written down above, say
a = abc. def. ghi. jkl, ¡3 — agj .bfi. cek. dhl;
and perform these in succession on the primitive arrangement il = abcdefghijkl. The
operation stands thus:
/Sail = fegkihlbjcda,
ail = bcaefdhigklj,
il = abcdefghijkl,
whence
/3a, = afhbeijcgl. dk,
is not a regular substitution; and, by what precedes, a, /3 cannot belong to a group.
But take the substitutions to be
a. (as before) = abc. def. ghi. jkl, /3 = ajg. bif. cek. dhl,
/Sail = iejkbhlfacdg,
ail = bcaefdhigklj,
il = abcdefghijkl,
then we have