Modified 2 Jan 2000
Horticultural Society of New York, 1902, pp. 12-15
Notes on Mendel's Methods of Cross-breeding
Charles C. Hurst, Burbage, Hinckley, Eng.

Mendel's Methods

(1) Single Characters

One of the most fruitful sources of confusion, in the older records of experiments in cross-breeding, has been the selection of the individual plant as the unit upon which to base the results.

The individual plant is made up of a large number of characters—organs, structures, whatever one may term them—distinctly marked off from one another, the points of difference both in form and in color being sometimes great and at other times small. In working out the inheritance of specific character in hybrid orchids in 1899, the writer became much impressed with the possibilities of variation in individuals, when a number of characters in each were considered together as one unit (Ref. 3: Report of the International Conference on Hybridization, London, 1899, in Jour. Roy. Hort. Soc., 1900, xxiv., pp. 106-117)

Some characters showed dominance of one parent, some of the other parent, while others were intermediate.

When these several variations occurred in twenty different characters, the possibilities of variations among the individual hybrids became very considerable, so much so that the results became quite unmanageable. Since that time the writer has been compelled to consider each single character on its own merits. It is true that, in come cases, the correlation of characters tends to modify this result to some extent, but in the case of the orchid hybrids in question the correlation was not very evident. From this experience it follows that in any statistics of inheritance a definite result can be determined by taking each single character as a distinct unit, completely ignoring, for the time being, the individual plant made up of many characters.

Mendel apparently was the first to see this clearly, and acted upon it in his experiments with Pisum, with remarkable results.

(2) Constant Characters

Next comes the important question of ancestry. From the earliest times it has been observed that in many instances offspring have resembled their grandparents or their more remote ancestors, rather than their actual parents. So that in experimental crossing, if two parents be chosen, each of whose ancestry is unknown or perhaps consists of complicated factors, the resulting offspring are either incomparable and incomprehensible, or they vary among themselves in bewildering confusion. The result, in any case, is chaos, and goes a long way to account for the many contradictory records which we find in the experiments carried out in the old style. Mendel, in his experiments, carefully and skilfully avoids this confusion by crossing together only constant and fixed races, i.e., each parent has been the product of repeated self-fertilization, so that its ancestry has been practically the same for many generations.

This effectually eliminates all the possible complications which might be caused by the influence of the immediate ancestors at any rate, though how far it affects the possible reversion to more remote ancestors is difficult to say. The writer, in his experiments with orchids, has chosen distinct species only as parents, and in this way, perhaps, reduces the possibilities of reversion still more. De Vries, Correns, Tschermak and Bateson have all for the most part followed or carried out Mendel's method by crossing constant races, and it is quite possible that some of their apparent exceptions to Mendel's results may have been due to their crossing particular races which were not really so fixed and constant as they believed them to be.

As we have seen, Mendel carefully avoided this by selecting in the first instance fixed parents of pure descent; these he further tested for two years, and satisfied himself as to their perfect constancy and fixity, and side by side with his crossing experiments he was careful to carry out "control" experiments with these parents by still further testing their constancy and fixity through all the generations.

It is just possible that these precautions of Mendel may explain the general uniformity of his results as compared with those of his disciples and some of his critics. (Ref. 4: cf. Weldon in Biometrika, 1902, i., pp. 228-254. For complete history, exposition and Bibliography of the Mendelian question, see Bateson's admirable hand-book on "Mendel's Principles of Heredity," Cambridge University Press, 1902.)

(3) Differential Characters

The third point worthy of note in the methods of Mendel is that the characters selected for crossing must not only be single and constant, but also differential in the two parents. If the single characters be nearly alike in the two parents it will be impossible to determine which parent the offspring resembles in that character, because all three would necessarily be nearly alike, i.e., the offspring and its two parents. On the other hand, the wider the difference between the pair of parental characters, the more clearly defined will be the single character in the resulting offspring, and consequently the easier it will be to refer the resemblance in the offspring to either parent.

Mendel, in his experiments, takes single constant characters in the parents which are distinctly differential and which can be clearly defined in the offspring.

(4) Dominant Characters

The fourth point in Mendel's methods is distinctly new, and that is the crossing together only of Dominant and Recessive characters, i.e., one of the characters of the differential pairs is always distinctly dominant over the other one, which latter is known as the Recessive character.

This serves a useful purpose in giving uniformity in the first generation, and thus avoids the great difficulty of working on to the next generation with results which are not uniform.

For instance, if the pair of characters were of fairly equal potency, they would, on the whole, be intermediate—either blended or mosaic, tending to one parent and the other alternately. It is obvious, therefore, that in a case like this, if one wished to carry on the experiment to further generations, the lack of uniformity in the first generation would complicate the experiment so much as to make it almost unworkable.

Mendel avoids this by the selection of Dominant and Recessive characters only, consequently his results can easily be recorded and tabulated statistically in all the generations.

(5) Large Numbers

The fifth point of note in Mendel's methods is his use of large numbers, and in this respect he was without doubt far in advance of his predecessors and contemporaries. In the older experiments, for the most part, only a few individuals of each cross were raised, and hence the range of variations apparent was either extreme or scarcely perceptible, according to chance and circumstances.

This no doubt accounts in some measure for the many contradictory results obtained by different experimenters at different times. Mendel avoided these difficulties by raising large numbers of individuals in each generation, and in that way practically gauged the total range of variation in each case.

(6) Many Generations

Now we come to the sixth and last method of Mendel to be noted here, and that is that he was not content to stop at the first generation or even the second, as so many of his predecessors were, but he in all cases carried on the experiments to the third and fourth generations and in some cases to the fifth and sixth generations. Mendel saw clearly that this was absolutely necessary, though at the same time the work must have been very laborious, and it illustrates once more tha thorough and painstaking methods by which Mendel overcame all obstacles in his pursuit of truth. [Ed. note: Mendel had a gardener!]

To sum up the methods of Mendel: Those hybridists who desire to follow in the footsteps of Mendel and his disciples and help to elucidate the baffling problems of inheritance will find it essential to select parents for the original cross which possess characters which are at once Single and Constant and Differential and Dominant, and they must also be take care to raise large numbers of individuals in many generations for observation and comparison. By these methods alone will definite results be obtained.


CybeRose note: Mendel's results are well known, and often misinterpreted. For instance, it is sometimes supposed that Mendel discovered or proved that heredity is a matter Dominant and Recessive traits, with "incomplete dominance" and "super dominance" being minor inconveniences of little theoretical relevance.

In the above article Hurst makes it perfectly clear that Mendel deliberately chose Dominant and Recessive traits out of the available possibilities because they simplified the collection of statistical data. No where did Mendel claim that heredity is solely a matter of Dominant and Recessive traits, nor did he offer any "proof" for an assertion he did not make. Later Mendelists looked for Dominant and Recessive traits, but likely as not were disappointed. "Blended" and "Mosiac" inheritance are widely observed, and can be statistically bewildering because they are so variable. One may observe that White x Yellow gives Cream, but that the Cream color varies from nearly white to practically yellow. Selection among the progeny for a few generations has given rise to distinct strains with different dominance. One strain produces White heterozygotes which, when mated with siblings of the same genotype, give 3 White : 1 Yellow. Just as Mendel described. However, in another derived strain the heterozygotes are yellow, and give the reverse proportions of 3 Yellow : 1 White. Again these are Mendelian ratios. Mendel selected traits which were observed to be Dominant and Recessive. He made no effort to determine the nature of dominance, and did not attempt to prove that dominance is a fixed quality of these traits in all cases. He merely used existing dominance as a convenient tool in his statistical study of inheritance.

Regarding large numbers: Hurst wrote "practically gauged the total range of variation", which is significant. There is evidence in Mendel's choice of a word that he combined results from different crosses to give the desired numbers, and to show free segregation where in fact there may have been linkage. Subsequent researchers have also taken this statistical liberty, which raises questions about what they have been ignoring. Jones wrote in his Selective Fertilization (1928) "...when deficient or excessive ratios are obtained, they have been obscured by the prevalent practice of compiling data from many individuals of different pedigree and in successive generations in order to obtain as large numbers as possible. This procedure serves its purpose well. The thoroughgoing Mendelianist seldom fails to obtain 'very good ratios.'" Alas, these "very good ratios" are therefore invalid.

It is pointless at this late date to condemn Mendel for making an unwarranted assumption in his otherwise valuable work. However, we should not expect Mendelian segregation for all pairs of contrasting traits in all cases. Nor should we assume that a trait which is dominant in one strain or cross will remain dominant when involved in other crosses, or when transferred to a different strain or species. Fisher made a strong case for accepting that the dominance of a trait is itself a selectable characteristic which may involve multiple genes. Other research has demonstrated that cytoplasmic components can also influence the expression of "Mendelian genes". A gene which is apparently not expressed at all in one variety may be expressed, and even dominant, in the cytoplasm of a different strain or species.