Under suitable experimental conditions, short denatured DNA fragments may be bound by high molecular weight denatured DNA immobilized in agar (‘long DNA’). The proportion of fragments bound depends in part upon the degree of taxonomic relationship between the organisms from which the two DNA components are taken. Before the method can be used as a measure of genetic relationship between closely related species or between intra-specific groups, its resolving power requires to be increased. The present study explores some modifications of the method from this point of view, using the degree of binding between DNA from laboratory mice and rats as a model.

Long rat DNA was less effective than long mouse DNA in binding short DNA fragments from either source. It was therefore necessary to test reciprocally, using both short mouse and short rat DNA on long DNA from each species. Combining reciprocals gives a measure of the proportion of fragments bound in within-species and in between-species combinations. With rats and mice, between-species binding was at about 70% of the within-species level. The discrepancy could be increased, and hence the discriminating power of the method improved, if the binding reaction was partially or totally ‘blocked’ with an excess of unlabelled DNA fragments in an appropriate species combination.

Even when short and long DNA came from the same source, only about a third of the fragments were bound. The bound and unbound fractions were recovered, and each reincubated with a fresh sample of long DNA. The two fractions then behaved very differently, suggesting that the original DNA preparation is heterogeneous. The initially unbound DNA showed a very low binding ability in subsequent incubations and may be considered virtually ‘unbindable’.

When the first incubation is a between-species one, the unbound fraction should include not only this ‘unbindable’ DNA, but also any DNA characteristic of that species alone, and not held in common by both species. If it could be isolated, this DNA fraction might be expected to discriminate maximally between the two species. By using three successive cycles of incubation, a fraction was obtained which showed considerably improved discriminating power, in that it showed only about 30% as much binding in between-species as in within-species combinations.