Below is a brief summary of the chief points of interest which our investigations appear to have suggested so far:

That the most convenient method of sampling roots for analysis is to remove a core from each root, and that when using this method at least 50 roots must be cored in order to obtain a sample representing the composition of the bulk of roots grown on a field.

That a large proportion of the commonly grown strains of mangels may be assigned to one or other of five types.

That of these types, four have their cropping power and percentage of dry matter so nearly in inverse proportion that they yield practically the same weight of dry matter per acre.

That the fifth type, Long Bed, yields considerably more dry matter per acre than the other four varieties.

That large roots on the average contain more water and less dry matter than smaller ones.

That there is a considerable variation in the composition of mangels from year to year, probably depending on such conditions as rainfall and sunshine at particular periods of growth.

That manurial treatment causes distinct variations in composition, the most noticeable point being that excessive applications of nitrogen delay ripening and decrease the percentage of dry matter.

That different farms grow roots of different composition.

That there is very great variation among individual roots of the same variety grown side by side, in content of dry matter, sugar, and nitrogen, and in size, shape, colour; in fact, in all the characters which we have been able to observe.

That there is so little correlation between the different characters that it is possible to pick out for seed-mothers large roots containing high percentages of dry matter rich in any desired constituent, and it is suggested that, from analogy with the sugar beet, continuous selection carried out in this manner may result in improvement in any desired direction.

That since colour, shape, and specific gravity of root or of juice are shown not to be correlated with percentage of dry matter, sugar, or nitrogen, selection for these characters is not likely to lead to any improvement.