406 
Journal of Agricultural Research voi. xxm, no. 6 
growth of the coccoids, which were always found immotile and Gram¬ 
positive. Figures 19 and 20 illustrate the upgrowth to larger globular 
cells, and figures 22 to 24 the streching to small and large non-sporulating 
and sporulating rods. The coccoids shown in figures 17 and 18 look very 
much like typical micrococci, and they may, in fact, assume a great 
stability of growth. But sooner or later they, too, will either return 
to the large cell form or elongate to regular or irregular rod forms. One 
of our strains (coccoids of Azotobacter agile , No. 7) remained fairly stable 
as a pink “micrococcus,” for not less than five years in 71 transfers, 
and then it returned to the large original growth type only if kept either 
in milk or in mannite-nitrate solution for one to two months. Milk 
and potato agar were usually most favorable for a return to the large 
cell type, while beef agar, beef broth, and mannite-nitrate solution in¬ 
duced the develpment of rodlike and fungoid forms. 
The dwarfed cell type, first discovered in an old culture of Azotobacter 
chroococcum ( 26 ), can be evolved from Azotobacter cultures as easily as 
or even more easily than the occcoid growth. It is either immotile and 
Gram-negative or (the smallest units) motile and Gram-positive. The 
liberation of these smallest specks of mostly nuclear material from the 
larger cells (by budding or by so-called granular decomposition) can be 
seen in every culture (fig. 3 and 5 of PL 1). But separate development 
is usually slow and inconspicuous, and the isolation of this type of growth, 
therefore, not always easy. Pure cultures once established, however, are 
much inclined to reproduce larger cells, especially rods and fungoid growth. 
Figures 25 to 36 on Plate 3 show the typical growth and the alterations 
most frequently observed. A close study of the minute forms character¬ 
istic of this stage (fig. 25 and 26) reveals the tendency to present a 
more or less irregular, often wedge-shaped appearance. This tendency 
becomes more noticeable when the upgrowth takes place either by a 
uniting of the small granular bodies in slimy threads (fig. 28 and 29) or 
by their stretching to small rodlike cells (fig. 30 and 31). On the other 
hand, very regular, weakly staining small ovals may appear, or the small 
irregular cells may assume globular shape (fig. 27), simulating in this 
case small micrococci. The latter process is always ascertainable in 
cultures several weeks old, whose miscroscopic appearance is very similar 
to that shown in figure 18 on Plate 2. Large rod-shaped, threadlike, 
or large globular cells are comparatively rare in cultures of this type; 
if they appear they are probably always produced by the symplasm 
(fig. 32 to 36). The simultaneous reproduction of small cells, of long 
slimy threads, which are occasionally branched, and of large ovals and 
globules deserve special attention (fig. 33 and 36; both pictures made from 
the same slide). The gradual upgrowth of rods from small to large size 
(shown in fig. 34) is equally of great interest. The pale, small ovals, 
mentioned above, seem to be able to assume directly the qualities of 
an endospore, able to reproduce a fairly large sporulating rod. This 
transformation was repeatedly observed with our cultures No. 1 and 2. 
The fungoid cell type is represented by figures 37 to 48 on Plate 4. 
Figures 37 to 39 illustrate that this irregular cell type may in fact assume 
distinctly fungoid appearance; figure 11 on PI. 1 deserves also renewed 
inspection. Figures 37 and 38 picture the cells most frequent and 
most typical; they are made up of unstainable slime and darkly staining 
granules of nuclear material. Temporarily this slime is soluble in boil¬ 
ing water, and therefore the fungoid type of growth was not recognized 
as such in our preliminary report (28). But from the unstainable 
