304 CARL HALLQVIST 

varieties of this type differing in the intensity and in the distribution 
of the colour; only the darker one of his types, which at the same time 
is the one generally found, has been used in my experiments, but seeds 
resembling the lighter, thin coloured types have often been found 
as modifications. 
The fourth type proposed by Kasanus has been used in one of 
my experiments (Pl. II, fig. 7). It lacks the pearl-gray ground-co- 
Jour; in its place the earth-brown marbling colour covers the whole 
surface of the seed with the exception of the white flecks. The pig- 
ment groups are found here only in two grades of density, sporadically 
in the white flecks and closely compacted in the earth-brown areas. 
The intermediate grades of density do not occur, and the gray colour, 
consequently, does not occur either (fig. 1 b). I have characterized 
this type as the earth-brown, not marbled. It is very easy to distinguish 
this type from the marbled as no transitions are found. 
The earth-brown seed colour is found correlated with blue, tinged 
blue, bluish red and tinged red flowers, but is not found in violet, pure 
red and white. 
The rust-brown seed colour, (Pl. II, fig. 9 and 10) due to 
another kind of pigment, is not mentioned in the literature. This type 
includes also marbled and non-marbled varieties. The marbled variety 
(Pl. II, fig. 10) has a gray field, almost cream-coloured, a rust-brown 
marbling and white flecks. Seeds from the earth-brown type are some- 
times modificatory rust-brown when poorly developed. If only sound 
and fully developed seeds are taken for comparison, no difficulty is 
met with in classifying the types. The rust-brown seed colour is found 
correlated with violet and pure red flowers. 
The non-marbled variety of the rust-brown type has been obtained 
as a ‘recombination product from my experiments. It also lacks the 
gray ground colour as does its earth-brown analogue. It may appear 
peculiar that the gray ground colour and the marbling colour of the 
three-coloured type depend on different quantities of the same pig- 
ment. The investigations of FruüuwIRTH and Kasanus as well as my 
own experiments bear out this fact, however, and it is further strength- 
ened by the observations made on the colour of seeds still immature 
and enclosed in the green pods. The gray as well as the marbled 
parts of the seed have the same colour at this stage, and the difference 
is wholly one. of intensity. In the earth-brown type there are two dif- 
ferent grades of intensity of blue, in the rust-brown type of violet. 
The white seed (Pl. II, fig. 11) has a faint rust-brown, hook-for- 
