AUB and his colleagues found that a contaminant in wheat germ lipase preparations agglutinated certain tumour cells, but not normal cells1,2. The agglutinating molecule was later purified by Burger and Goldberg3 and characterized as a plant lectin (wheat germ agglutinin) that binds to specific terminal oligosaccharide residues containing N-acetyl-D-glucosamine. The ability to agglutinate tumour cells at concentrations which do not affect normal cells has been found for several other lectins with different saccharide binding specificities: concanavalin A4,5 (Con A), soy bean agglutinin6, Ricinus communis agglutinin7, Pisum sativum agglutinin8 and others9. Burger proposed that the difference in agglutinability of tumour cells was a consequence of the exposure of “cryptic” lectin binding sites during transformation3,10. Several investigators have found, however, that there is essentially no difference in the number of lectin sites on normal and transformed cells11–15. Another explanation for the differential agglutinability of tumour cells was that the distribution of lectin sites at the cell surface could change to a topographic (“clustered”) state which is more favourable for cell agglutination16–18. Experimental evidence for a topographic difference in the distribution of Con A sites on transformed, compared with normal, fibroblast cell surfaces has been obtained using ferritin-conjugated Con A16 and peroxidase-bound Con A19,20. In these studies mounted cell membranes16 or intact cells19,20 were labelled at room temperature with the Con A reagents. The mounted cell membranes were observed directly to obtain the topographic distribution of ferritin-Con A sites16; or the intact cells19,20 were treated with Con A and peroxidase, fixed, dehydrated and embedded in plastic for thin section observation of an electron-dense peroxidase product at the cell surface19,20. By either method the Con A membrane sites on tumour cells were distributed in a more clustered state than in normal cells16,19,20. The clustered Con A sites seem to be involved in cell agglutination, because trypsinized normal cells (which are agglutinable) also show a similar reorganization of Con A sites18. When trypsinized 3T3 cells were agglutinated with ferritin-Con A and the cell aggregates removed, fixed, embedded in ‘Epson’ 812, sectioned and their cell membranes examined, the ferritin-Con A was found in clusters at the sites of contact between the agglutinated cells18.