nature of the change. For experimental purposes a single mutation does
not suffice; it must be studied repeatedly, and be produced more or less
arbitrarily, according to the nature of the problems to be solved. And
in order to do this, it is evidently not enough to have in hand the
mutated individual, but it is indispensable to have also the mutable
parents, or the mutable strain from which it sprang.

All conditions previous to the mutation are to be considered as of far
higher importance than all those subsequent to it.

Now mutations come unexpectedly, and if the ancestry of an accidental
mutation is to be known, it is of course necessary to keep accounts of
all the strains cultivated. It is evident that the required knowledge
concerning the ancestry of a supposed mutation, must necessarily nearly
all be acquired from the plants in the experimental garden.

Obviously this rule is as simple in theory, as it is difficult to carry
out in practice. First of all comes the book-keeping. The parents,
grandparents and previous ancestors must be known individually. Accounts
of them must be kept under two headings. A full description of their
individual character and peculiarities must always be available on the
one hand, and on the other, all facts concerning their hereditary [24]
qualities. These are to be deduced from the composition of the progeny,
and in order to obtain complete evidence on this point, two successive
generations are often required. The investigation must ascertain the
average condition of this offspring and the occurrence of any deviating
specimens, and for both purposes it is necessary to cultivate them in
relatively large numbers. It is obvious that, properly speaking, the
whole family of a mutated individual, including all its nearer and more
remote relatives, should be known and recorded.