Here we meet with a great difficulty. Hitherto varieties and subspecies have never been clearly defined, or when they have been, it was [225] not by physiological, but only by morphological research. And the claims of these two great lines of inquiry are obviously very diverging. Morphological or comparative studies need a material standard, by which it may be readily decided whether certain groups of animals and plants are to be described or de-nominated as species, as subspecies or as varieties. To get at the inner nature of the differences is in most cases impossible, but a decision must be made. The physiological line of inquiry has more time at its disposal; it calls for no haste. Its experiments ordinarily cover years, and a conclusion is only to be reached after long and often weary trials. There is no making a decision on any matter until all doubtful points have been cleared up. Of course, large groups of facts remain uncertain, awaiting a closer inquiry, and the teacher is constrained to rely on the few known instances of thoroughly investigated cases. These alone are safe guides, and it seems far better to trust to them and to make use of them for the construction of sharp conceptions, which may help us to point out the lines of inquiry which are still open. Leaving aside all such divisions and definitions, as were stamped with the name of provisional species and varieties by the great systematist, [226] Alphonse De Candolle, we may now try to give the proofs of our assertion, by using only those instances that have been thoroughly tested in every way. We may at once proceed to the retrogressive or negative varieties. The arguments for the assumption that elementary species owe their origin to the acquisition of new qualities may well be left for later lectures when we shall deal with the experimental proofs in this matter. There are three larger groups of facts, on which the assumption of latent characters in ordinary varieties rests. These are true atavism, incomplete loss of characters, and systematic affinity. Before dealing with each of these separately, it may be as well to recall once more that in former lectures we have treated the apparent losses only as modifications in a negative way, without contemplating the underlying causes.