There are conflicting views on the nature of subgroups A, and A2 of blood group A, One holds that the same determinants are present on either, but that there are fewer determinants on A2 than on A, erythrocytes. Soluble A2 and A, substances would thus have the same kinds of determinants, but in different numbers. The A1 and A2 transferases are different enzymes, but the A2 enzyme is less efficient than the A, transferase, but it has the same specificity (1) in adding terminal N-acetyl-D-galactosaminyl residues to precursor blood group H oligosaccharide side chains. Such a difference in enzymatic activity has been proposed to be responsible for the H activity of A2 cells (2). A population of those anti-A, antibodies which do not agglutinate A2 erythrocytes is known and has been prepared from the purified IgM fraction of anti-A sera by absorption with A2 erythrocytes, but not from the IgG fractions of the same sera (3). Such anti-A antibodies have been assumed (3) to have a low affinity so that if they use only one of their ten valences per erythrocyte, they would be unable to hold two erythmcytes together. The anti-A, antibodies are hypothesized to agglutinate A, erythrocytes which have receptors that are more numerous and closer together, so that each erythrocyte could be bound by two or more of the combining sites of each antibody molecule.The other concept, based predominantly on immunochemical studies with A1 and A2 glycoproteins, favors a qualitative difference (4). Absorption with insolubilized polyleucyl A2 substance did not remove all of the anti-A, but left anti-A~; had all determinants been present on A2 as well as on A~ substances, no anti-A~ should have remained.Since several different determinants on soluble blood group A substances have been found (5, 6, 7), a basis for a structural difference between A~ and A2 exists. All A determinants have the structure ~ucal 2 DGalNAcal--*3DGal but differ in that this trisaccharide may be linked /~1--.3 or B1~4 to vGlcNAc to give