A specimen from the late Devonian Katsberg beds of Delaware County, New York, comprising a pyritized axis determined asCallixylon and leaves determined asArchaeopteris, is described and illustrated. The leaves are probablyA. macilenta Lesquereux and the axisC. zalesskyi Arnold, but because of doubt of the specific identity of the leaves a nomenclature transfer is delayed. Archaeopteris, the valid name for the plant represented by the organ generaArchaeopteris andCallixylon, is included withPitys andArchaeopitys in the Pityales which, together with the orders Aneurophytales and Protopityales, is assigned to the newly recognized class Progymnospermopsida. This class includes woody, pteridophytic plants bearing (where known) large compound leaves or leaf-like branch systems. In numerous characters of both external morphology and internal structure these plants are remarkably similar to two groups of gymnosperms, the Pteridospermales and Cordaitales, which are, respectively, the most primitive groups of the cycadophyte and coniferophyte lines of gymnosperm evolution. Because the Progymnospermopsida are pteridophytic they can-not be logically classified with the ovule-bearing gymnosperms, but it is very likely that they comprise the ancestral complex from which the major groups of gymnosperms evolved. Certain primitive features, especially of the Aneurophytales, suggest that the Progymnospermopsida are descended directly from some psilophyte-like ancestors. The existence of such a group of pteridophytic plants, of possible psilophytic origin, obviously not ferns, showing unmistakeable gymnosperm characters, and which preceded any known gymnosperms in time, eliminates the necessity to consider any group of ferns, known or unknown, as ancestors of the gymnosperms. This supports the separation by Bold (1956) of the Filicineae and Gymnospermae, and in part his abandonment of Pteropsida. On the other hand it suggests that the major groups of gymnosperms (possibly excluding the Gnetales) have a common ancestry, are consequently genetically related, and should, therefore, be retained in a single inclusive taxon.