As no two plants are alike in outward appearance, so it is certain they must differ inwardly, that is, the forces which determine the outward characteristics must be more or less differentiated in each plant. Moreover, every plant is made up of, or contains, a certain number of distinct characters, which are, of course, not interchangeable; for instance, it may be tall, have broad leaves, and red flowers, and so on. Each of these characters however has a certain range of variation, and forms one of a series or category of minor differences within that range which are interchangeable. Thus, the different colours or degrees of colour of two varieties are interchangeable, likewise the varying degrees of height, and so on. When, therefore, two species or varieties, containing one or more pairs of such differential characters, are crossed, there must necessarily be a conflict between the forces which determine the ultimate characteristics of the offspring. In some cases, e.g., where "continuous" variations are concerned, they will blend; in other cases they will not, and one of the two characters is then selected, and appears in the progeny to the total exclusion of the other. Or again, the incompatibility may be so extreme, or may perhaps extend to so many of the characters, that the determining forces of the germ plasm are unequal to the work of selecting and building up the structure, and this probably is the main consideration that sets a limit to the intercrossing of species. We may thus realise how the union of two dissimilar parents—that is two plants having differential interchangeable characteristics-may give rise to offspring with a considerable range of variation. Each seedling will be the result of a certain combination and compromise between the competing characters of the parents, and among the possible combinations individuals may arise which, outwardly at least, will appear quite unlike either parent, and so form the starting-point of a new variety, race, or species.

So far we have seen how variations in plants may be induced by cultivation, or change of conditions; and though as yet we

do not know how to produce any particular variation, we have seen how such as may arise can be accentuated and combined by cross-fertilisation and hybridisation. But the production of a variety is not always the ultimate object of the plant-breeder : he may desire to perpetuate it, to fix it, so that it will breed true, or to use it as the foundation of a more or less distinct strain, and he has then to deal with the problems of inheritance. Hitherto the only established principle of heredity on which we could rely was the universal experience that "like begets like," -that the offspring tend to resemble their parents—with its corollary, the law of regression to a mean, that is, that the descendants which depart from the type in one generation tend to revert towards it in subsequent generations. When therefore we are dealing with fixed races, or stable and constant varieties, and either self-fertilisation is practicable, or two closely similar individuals can be mated, there is no difficulty in perpetuating any characteristic, and improving and fixing it, by careful selection. But when breaking new ground and endeavouring to start new strains by combining dissimilar characteristics, or when dealing with unstable hybrid and cross-bred races, plantbreeders have hitherto worked largely in the dark. Until the recent rediscovery of Mendel's experiments they have had no other guidance than their own incomplete experience, and have been compelled to rely chiefly on large numbers and on careful selection.

Considering the importance of these principles of inheritance which are now known as Mendel's Laws, to the student of heredity as well as to the practical breeder, it is strange that, though formulated more than forty years ago, they should not have been appreciated by his contemporaries, and should have remained practically unknown until their rediscovery and publication in 1900. In the short space of time that has elapsed since then they have been confirmed and amplified, and further investigations on similar lines are at present being instituted into other as yet unsolved problems of heredity.

Mendel's main discovery was of the definite proportions in

which the characters of the parents are distributed among their descendants. To take the simplest case, and ignoring for the moment all disturbing factors-when hybrids or cross-breds resulting from the union of parents, having each one of a pair of differential interchangeable characters, are self-fertilised, of their offspring some will contain one only of the characters of the original parents, some will be intermediate or will contain both characters, and some will contain the other character only; and the numbers of the individuals composing these three groups will be in a definite numerical ratio of 1: 2: 1. Thus for a pair of simple characters A and B, of every hundred seedlings twenty-five will have the character A, fifty will be "intermediate," and twenty-five will have the character B. This is Mendel's "Law of Segregation," or of the distribution among the offspring of the characters of the parents.

If individuals from these three groups are selected or "extracted" and again self-fertilised, those which contain either one or the other only of the original pair of characters will respectively transmit that character pure to their offspring, i.e., they will breed true to that character, and their descendants will continue to do so in every subsequent generation. On the other hand, the offspring of the "intermediates" will again split up into three groups as before, which will behave in future. generations as in the first, i.e., "intermediates" of every generation will always produce offspring half of which will be similar "intermediates," while the other half will be composed of equal numbers containing one or other of the characters respectively, and which will breed true to that character. This is Mendel's "Law of Purity."

When working with more than one pair of characters the results are, of course, more complex, but by applying the laws, the proportionate distribution among the offspring of the two or more characters taken together, and the subsequent course of their breeding can be readily calculated, and will always be found to agree with the actual results, if sufficient numbers are employed. There are furthermore some characters which are

apparently of a compound nature, and though on inbreeding they may act as simple characters, when crossed they split up into their component characters, which then form other combinations. From such cases, however, as have so far been analysed, there is no doubt that they follow the same laws, but the results are so complex that it is impossible to do more than allude to them here.

Whether these laws are of universal application time and further investigation will show; in their present form they have as yet been established only for discontinuous characters.

There is, however, one important phenomenon, by which the working of these laws is frequently obscured that of Dominance. This phenomenon of dominance seems to be very generally associated with discontinuous characters, and as a matter of fact Mendel in his experiments, working chiefly with fixed races of Peas (Pisum Sativum), deliberately chose such characters. In these cases the seedlings of the first cross of the original parents exhibit one of the two characters only— the dominant one. These when "selfed" produce offspring which exhibit the respective characters in the proportion of three dominants to one recessive. The recessives breed true from this generation, while further breeding shows that the dominants are composed, in the proportion of one to two, of pure dominants (which thereafter breed true to the dominant character), and impure dominants which contain the recessive character also, though latent (and which split up in future generations in the same proportion of three dominants to one recessive). It is shown, therefore, that, though one character is latent, the actual ratio is one pure dominant to two impure or hybrid dominants to one recessive, or D: 2 DR: R, as required by the general law. What is the cause of this phenomenon of dominance we do not know; it may be connected with the phylogenetic origin of the character, or it may be due to molecular structure or composition in some way akin to the atomic structure of stable chemical compounds. It is an interesting problem awaiting solution.

Of other disturbing factors I will only mention one-that of correlation of characters, as it is, perhaps, of all the unsolved problems of heredity of most importance to plant-breeders. At present our knowlege of correlation consists only of a scattered mass of unconnected and often contradictory facts, and the breeder, when introducing a new character into his strain, has no idea, beyond his immediate experience, whether it will carry other characters with it or not. For all he can tell it is as likely to be correlated with an undesirable character as not; or it may carry a character with it in one combination and not in another. If by the study of such authentic cases as are on record, and by further experiments, some general law could be evolved which would co-ordinate the facts and provide some guidance in future, apart from its scientific interest, it would undoubtedly be of the greatest value to the practical breeder. Here, as with the phenomenon of dominance, is an almost virgin field for the labours of the scientific horticulturist.

Beside these laws of heredity, which are, of course, the expression of the actual observations, there are furthermore certain theoretical conclusions that can be drawn from them, some of which are of considerable importance to the plantbreeder. From the invariable numerical proportions in which the characters of the original parents appear in their descendants, it follows that, though a plant or individual may contain more than one differential character of a category, the germ-cells, or "gametes "at any rate at the period of union-carry only one; and that, therefore, each germ-cell is pure for one such character. Consequently, when a plant contains two or (as in the case of compound characters) several characters of a category, it follows that these characters must segregate, or be distributed in equal numbers among the germ-cells; i.c., if the plant contains two characters, half the gametes will carry one, and half the other character, and so on. These are Mendel's theories of gametic purity and segregation.

The practical value of these laws and theoretical conclusions is obvious. When once a variation has been obtained No. 70. XXIV. 1.-JULY 1906