essentials in this species of improvement should always accompany each
other; for without good resources of keeping, it would be vain to
attempt supporting a valuable stock. This is true with regard to the
original stock. It is yet more evident when animals are absurdly
brought from a better to a poorer soil. The original stock will
deteriorate if neglected and half-starved, and the improved breed will
lose ground even more rapidly, and to a far greater extent."

A very brief resume of the preceding remarks may be expressed as
follows:

The Law of Similarity teaches us to select animals for breeding which
possess the desired forms and qualities in the greatest perfection and
best combination.

Regard should be had not only to the more obvious characteristics, but
also to such hereditary traits and tendencies as may be hidden from
cursory observation and demand careful and thorough investigation.

From the hereditary nature of all characteristics, whether good or
bad, we learn the importance of having all desirable qualities and
properties _thoroughly inbred_; or, in other words, so firmly fixed in
each generation, that the next is warrantably certain to present
nothing worse,--that no ill results follow from breeding back towards
some inferior ancestor,--that all undesirable traits or points be, so
far as possible, _bred out_.

So important is this consideration, that in practice, it is decidedly
preferable to employ a male of ordinary external appearance, provided
his ancestry be all which is desired, rather than a grade or
cross-bred animal, although the latter be greatly his superior in
personal beauty.

A knowledge of the Law of Divergence teaches us to avoid, for breeding
purposes, such animals as exhibit variations unfavorable to the
purpose in view; and to endeavor to perpetuate every real improvement
gained; also to secure as far as practicable, the conditions
necessary to induce or to perpetuate any improvement, such as general
treatment, food, climate, habit, &c.

Where the parents do not possess the perfection desired, selections
for coupling should be made with critical reference to correcting the
faults or deficiencies of one by corresponding excellence in the
other.

But to correct defects too much must not be attempted at once. Pairing
those very unlike, oftener results in loss than in gain. Mating a
horse for speed with a draft mare, will more likely beget progeny good
for neither, than for b