In Dr. Carpenter's "Introduction to the Study of the Foraminifera," he states that "_there is not a single specimen of plant or animal of which the range of variation has been studied by the collocation and comparison of so large a number of specimens as have passed under the review of Messrs. Williamson, Parker, Rupert Jones, and myself, in our studies of the types of this group_;" and the result of this extended comparison of specimens is stated to be, "_The range of variation is so great among the Foraminifera as to include not merely those differential characters which have been usually accounted_ SPECIFIC, _but also those upon which the greater part of the_ GENERA _of this group have been founded, and even in some instances those of its_ ORDERS" (Foraminifera, Preface, x). Yet this same group had been divided by D'Orbigny and other authors into a number of clearly defined _families_, _genera_, and _species_, which these careful and conscientious researches have shown to have been almost all founded on incomplete knowledge. Professor DeCandolle has recently given the results of an extensive review of the species of Cupuliferae. He finds that the best-known species of oaks are those which produce most varieties and subvarieties; that they are often surrounded by provisional species; and, with the fullest materials at his command, two-thirds of the species he considers more or less doubtful. His general conclusion is, that "_in botany the lowest series of groups,_ SUBVARIETIES, VARIETIES, _and_ RACES _are very badly limited; these can be grouped into_ SPECIES _a little less vaguely limited, which again can be formed into sufficiently precise_ GENERA." This general conclusion is entirely objected to by the writer of the article in the "Natural History Review," who, however, does not deny its applicability to the particular order under discussion, while this very difference of opinion is another proof that difficulties in the determination of species do not, any more than in the higher groups, vanish with increasing materials and more accurate research. Another striking example of the same kind is seen in the genera Rubus and Rosa, adduced by Mr. Darwin himself; for though the amplest materials exist for a knowledge of these groups, and the most careful research has been bestowed upon them, yet the various species have not thereby been accurately limited and defined so as to satisfy the majority of botanists. In Mr. Baker's revision of the British Roses, just published by the Linnaean Society, the author includes under the single species Rosa canina, no less than twenty-eight named _varieties_, distinguished by more or less constant characters and often confined to special localities; and to these are referred about seventy of the _species_ of Continental and British botanists.