ut is believed to be from a cross. Mr. Salter, of
Hammersmith, states[838] that he has himself known this purple variety
to produce the lilac, the rose-crimson or _conspicuum_, and the red or
_coccineum_ varieties; the latter has also produced the _rose d'amour_;
so that altogether four varieties have originated by bud variation from
Rollisson's Unique. Mr. Salter remarks that these four varieties "may
now be considered as fixed, although they occasionally produce flowers
of the original colour. This year _coccineum_ has pushed flowers of
three different colours, red, rose, and lilac, upon the same truss, and
upon other trusses are flowers half red and half lilac." Besides these
four varieties, two other scarlet Uniques are known to exist, both of
which occasionally produce lilac flowers identical with Rollisson's
Unique;[839] but one at least of these did not arise through
bud-variation, but is believed to be a seedling from Rollisson's
Unique.[840] There are, also, in the trade[841] two other slightly
different varieties, of unknown origin, of Rollisson's Unique: so that
altogether we have a curiously complex case {379} of variation both by
buds and seeds.[842] An English wild plant, the _Geranium pratense_,
when cultivated in a garden, has been seen to produce on the same plant
both blue and white, and striped blue and white flowers.[843]

_Chrysanthemum._--This plant frequently sports, both by its lateral
branches and occasionally by suckers. A seedling raised by Mr. Salter
has produced by bud-variation six distinct sorts, five different in
colour and one in foliage, all of which are now fixed.[844] The
varieties which were first introduced from China were so excessively
variable, "that it was extremely difficult to tell which was the
original colour of the variety, and which was the sport." The same
plant would produce one year only buff-coloured, and next year only
rose-coloured flowers; and then would change again, or produce at the
same time flowers of both colours. These fluctuating varieties are now
all lost, and, when a branch sports into a new variety, it can
generally be propagated and kept true; but, as Mr. Salter remarks,
"every sport should be thoroughly tested in different soils before it
can be really considered as fixed, as many have been known to run back
when planted