A survey of root meristems shows that in general, closed meristems, those with separate regions generating outer and inner tissues, derive the epidermis with the cortex in monocotyledons and with the cap in dicotyledons. Open meristems, where the cells between the stelar pole and cap divide transiently producing an unstable boundary between the cap and the rest of the root, also show a distinction between the classes, monocotyledons displaying greatest affinity between epidermis and cortex, dicotyledons between epidermis and cap.Exceptions in monocotyledons occur where there is a discrete epidermal meristem, an anticlinally di\iding, monostromatic sheet of cells between cortex and cap. This condition has been reliably found only in a few Boating plants and it is associated with the root being derived from both pericycle and endadermis of the mother organ. This is uncomnKin as a permanent feature In angiospcrms though not confined to those with discrete epidermises. The mother tissue producing the epidermis depends on the species. In many other monocotyledons with closed meristems the epidermis is functionally discrete, for the branching of the epidermal cell files from the cortical files is seen within the quiescent centre, i,e, it occurred during the meristem's initiation.Exceptions in dicotyledons nccur in the Xymphaeales. In their closed meristems the epidermis is part nf the cortical complex and in their open meristems its affinity with the cortex is manifest by early separation of the cap creating a cylindrical cleft in some genera. Differentiation of epidermal trichoblasts is also by a mode more characteristic of monocotyledons. Nelumbo roots, previously described as lacking an epidermis, as in gymnosperms, have closed menstems of dicotyledonous type and epidermises vvithoui distinctive trichobiasts, supporting Nelumbo's transfer fronn the Nymphaeales.The survey has corrected many errors of observation and interpretation in the literature enabling better use to be made of the mode of origin of r(K)t epidermises in discussing e\'olution and the new reel as si fi cation of angiosperms.