Normal spleen cells were separated in dishes coated with thin layers of DNP-gelatin or NIP-gelatin into binding and nonbinding cells and stimulated in vitro with DNP- and/or NIP-conjugated polymerized flagellin (POL). Hapten-specific unresponsiveness was induced in the binding cell population by melting the gel at 37 degrees C or in unfractionated cells by pretreatment with soluble hapten-gelatin and could be reversed by treatment with collagenase. A specific enrichment of anti-DNP and anti-NIP antibody-forming cell precursors (AFCP) could be demonstrated in the binding cell populations after treatment with collagenase in cultures with or without "feeder" cells. However, the response of small numbers of unfractionated and purified hapten-specific spleen cells was suboptimal even in the presence of mitomycin-treated or irradiated feeder cells. Optimal numbers of anti-DNP (anti-NIP) antibody-forming cells were generated by small numbers of normal or purified spleen cells in the presence of spleen cells depleted of anti-DNP (anti-NIP) AFCP. In this system the response of only 2 times 10-4 purified hapten-specific cells was higher than the response of 10-6 unfractionated cells. Purified DNP-specific cells responded only to DNP-POL but not to NIP-POL and purified NIP-specific cells responded only to NIP-POL but not to DNP-POL. The degree of enrichment of anti-DNP AFCP decreased with increasing numbers of binding cells. NIP3-gelatin layers bound four to five times less spleen cells than DNP2-gelatin layers and the enrichment of anti-NIP AFCP (about 300-fold) was three times greater than the enrichment of anti-DNP AFCP (about 100-fold). The immunological significance of hapten-gelatin binding cells which apparently failed to respond to antigen is discussed.