Nonacid plasma glycolipids from Lewis-negative individuals of nonsecretor, partial-secretor and secretor phenotypes were prepared and separated by thin- layer chromatography and immunostained with radiolabelled Lewis antibodies. Lewis-positive plasma and intestinal epithelial cell glycolipids from Caucasians representing the four recognized Lewis and secretor combined phenotypes were used as controls. By presenting these purified total glycolipids in a cell-free environment to Lewis antibodies we were able to demonstrate the presence of small amounts of Lewis antigens in Lewis-negative individuals. It is shown that lacto- tetraosylceramide and extended precursor glycolipids are present in all Le(a-b -) nonsecretors. Le^b was detected in 1 of the 3 Le(a-b-) nonsecretor plasmas and in the intestinal sample of the same phenotype. Lactotetraosylceramide was absent but H type 1 and Leb were both present in all group O Le(a-b-) secretors, and extended H typel reactive structures were also found in the partial secretor. These results clearly demonstrate that although the Lewis-negative phenotype exists at the serological level, this phenotype is not an ‘all-or-nothing’ phenomenon at the chemical level. We also show that in the presence of reduced fucosyltransferase activity, increased elongation of the precursor chain occurs, which allows us to postulate that fucosylation of the precursor prevents or at least markedly reduces chain elongation.

DiscussionThe properties necessary for the acid alcohol fastness of tubercle bacilli are an intact cell or the lipoidal portion of the cell mycolic acid or both.5Nyka67 has previously shown that periodic acid oxidation of tubercle bacilli during fixation and on paraffin sections caused them to stain strongly by the methenamine silver and Ziehl-Neelsen techniques and that without oxidation the stain was poor or negative. Whether the stain is intensified by the oxidation of the ethylene groups of the organisms' unsaturated lipids8 or by the oxidation of the polysaccharide constituents of the bacillus to form aldehydes stainable by basic fuchsin is not known.In this experiment, which was performed with epoxy resin sections, the same results occurred. It was also found that bromination for longer than 60 s produced a negative stain which could not be reversed by oxidation. Bromination for 30-60 s and subsequent oxidation produced a positive result. It follows that it is essential the tissue remains in the bromine vapour for the shortest possible effective time.It seems probable that mycolic acid is extracted by the presently used resin solvent ethanolic sodium hydroxide or blocked in the case of bromine.9 It 111 seems probable that the positive result after oxidation is dependent on the structure of the intact cell rather than the mycolic acid. I am grateful for the excellent technical assistance of J Kan and J Holzl. References 'Chang SC. Haematoxylin-eosin staining of plastic-embedded tissue sections. Arch Pathol 1972; 93:344-51. 2 Aparicio SR, Marsden P. Application of standard microanatomical staining methods to epoxy resin-embedded sections. J Clin Pathol 1969;22:589-92. Lane BP, Europa DL. Differential staining of ultra thin sections of epon-embedded tissues for light microscopy. J Histochem Cytochem 1965; 13:579-82. 4Yenson J. Removal of epoxy resin from histological sections following halogenation. Stain Technol 1968;43:344-6. Berg JW. Acid-fastness as a histochemical test. J Histochem Cytochem 1953;1:436-41. 6 Nyka W. Studies on Mycobacterium tuberculosis in lesions of the human lung. A new method of staining tubercle bacilli in tissue sections. Am Rev Resp Dis 1963;88:670-9. Nyka W. Method for staining both acid-fast and chromophobic tubercle bacilli with carbolfuchsin. J Bacteriol 1967;93: 1458-60. Wolman M. The lipids stained by the periodic acid-Schiff technic. Stain Technol 1956;31:241-5. Harada K. The nature of mycobacterial acid-fastness. Stain