Dextran fractions from NRRL strain Streptococcus sp. B-1526 and the native, structurally homogeneous dextrans from Acetobacter capsulatum B-1225, Leuconostoc mesenteroides B-1307, and L. dextranicum B-1420 were examined by 13C-n.m.r. spectroscopy at 90°. Dextran B-1526 fraction I and dextran B-1420 were also examined by g.l.c:-m.s., methylation-structural analysis. All of these dextrans and dextran fractions branch, either primarily or exclusively, through α- d-(1→4)-glucopyranosyl linkages; however, their degrees of branching differ. Several 13C-n.m.r. resonances that are diagnostic for 4,6-di- O-substituted α- d-glucopyranosyl residues have been identified. Comparison was made with dextrans from L. mesenteroides B-742 fraction L and Streptobacterium dextranicum B-1254 fraction S[L], for which previously published, methylation-structural analyses had established the presence of 4,6-di- O-substituted α- d-glucopyranosyl residues at the branch points. These fermentation culture, and in a sedimented gum-phase (fraction I). The product from the soluble phase is designated here as fraction S in order to simplify the terminology. Originally 7, this product was not designated a fraction, because it was, by definition 8, the main dextran product. The same distinction also applies to the pairs of products from strains B-1380, B-1420, and b-1394 (see ref. 7). The attempts thus made to establish the significance of the phase separation were indeterminant. Methods.— Methods previously described were used for the mythylation 9 of the dextrans and for structural analysis 6.38 by combined g.l.c-electron-impact mass spectrometry of the aldononitriles. For each permethylation, three successive Hakomori 39 methylations were employed on an initial, 40-mg sample, with ∼80% (final weight) recovery of each permethylated dextran. Successive formolysis and acetic acid hydrolysis were employed, and, after each step, the resulting solutions were clear, colorless, and free from suspended material. All mass spectra were recorded with a Hewlett-Packard 5980A GC/MS integrated g.l.c.-m.s.-computer system. The g.l.c. peak-integrals reported in Table II were obtained with a Barber-Coleman Series 5000 g.l.c. instrument equipped with hydrogen-flame detectors. On-column injection with glass columns (2mmi.d. x 1.23m) was employed for all chromatograhy. The 13C-n.m.r. conditions and the methods for the preparation of dextran samples have been described 4. In general, a Varian XL-100-15 spectrometer equipped with a Nicolet TT-100 system was employed in the Fourier-transform mode. The dextran samples, ∼0.3g/4 mL of deuterium oxide, were maintained at 90°. Chemical shifts are expressed in p.p.m. relative to external tetramethylsilane, but were actually calculated by reference to the solvent lock-signal. The convolution-difference resolution-enhancement (c.d.r.e.) technique has been described 40.