416 
Journal of Agricultural Research 
Vol. XXIII, No. 6 
of A. Beijerinckii were also found to be able to develop a dark brown or 
black color when they were kept for several months in tubes containing 
mannite-nitrate solution and partially immersed strips of filter paper. 
Directly above the solution the characteristic thick yellowish to brownish 
slime was always noticeable, but at the upper edge of the paper strips 
after four to six weeks a thin dark brown to black growth became visible, 
and the microscopic inspection revealed the fact that the cells in this case 
were much more solid and more uniformly stained than those making the 
more slimy light-colored growth. The latter are always granular and 
only partially stainable, as is characteristic of A. Beijerinckii. With 
A. agile analogous observations were made. The slimy growth was 
again of light brown color like that of A. Beijerinckii , and at the upper 
edge of the paper the same black development appeared in old cultures. 
If A . agile does not produce its characteristic green pigment, as frequently 
happens, its growth is hardly distinguishable from that of A. Beijerinckii; 
but here other types of growth as well as slight differences in morphology 
permit a clear differentiation, whereas there are no such differences 
between A . Beijerinckii and A . chroococcum. A . vitreum has never shown 
any pigmentation of its large nonsporulating cells, which practically 
always retained their typical globular shape. The only exception 
noticed thus far was discussed in another paper ( 26 ); the large rodlike 
forms seen in this case were inclined to assume endosporulation, which 
could not be fully established, however. The large globular cells of 
A. vitreum leave no doubt about their being gonidangia or microcysts, 
and their behavior and appearance is very similar to that of globular 
gonidangia and microcysts of A. agile . A comparison of figure 6 on 
Plate A of our preliminary communication ( 28 ) with figure 94 on Plate, 
8 of the present paper may illustrate this similarity. These large cells 
are Gram-negative, as is typical for the nonsporulating large cells of 
A. chroococcum , A. Beijerinckii , and A . agile, but the somewhat smaller 
globular cells of A . vitreum , which often occur in sarcina formation, are 
Gram-positive. In this as well as in all other respects they display the 
character of the coccoid growth common to all Azotobacter strains; in 
fact, with all of the latter cultures microscopic pictures were obtained, 
especially from potato agar, which looked exactly like A. vitreum . In 
the first Azotobacter paper ( 30 ) it was mentioned that a white sarcina 
was grown from A. agile, as was a yellow sarcina from A. Beijerinckii . 
No explanation could be given at that time; but to-day we know that 
it was the regenerative bodies we had before us, and we are now also able 
to show that A. vitreum may grow in other types equal to those of 
A. agile —that is, A. vitreum is a variety of A. agile , as is A. Beijerinckii 
of A. chroococcum. Cultivation of the large nonsporulating cells in 
slightly acid mannite-nitrate agar (P H 6.o) always gave during the first 
two weeks good typical growth with strong characteristic pigmentation; 
in older cultures, however, the tendency to produce small cells became 
very marked and was repeatedly used to great advantage. Potato agar, 
on the other hand, favored especially the coccoid growth, while in soil 
quick changes to all different cell types were observed simultaneously. 
2 .—COCCOID GROWTH 
As was explained before, this type of growth results when regenera¬ 
tive bodies cease to act as such and begin to multiply by fission and by 
budding. It is self-evident that in the beginning such strains are often 
