SIR -We are grateful to Edwards et al. (this issue) for a preview of their paper in which they object to the nomenclature and classification of lithostratigraphic units that we have employed to describe the Mesozoic sequence in Svalbard. They present an alternative scheme in which they also take note of new data. In certain respects, therefore, we are bound to accept their suggestions, but in some others we argue below that their criticism is not well based.In parts of this discussion there is clearly room for more than one opinion; we debated between ourselves what descriptive scheme we should employ at the time of our publication (1975-6), and we now argue that in the interests of stability a scheme proposed as part of a new and systematic description should stand until an altogether more thorough study supersedes it. This has been policy since our group's first substantial publication on the Mesozoic geology of the region in 1965 (Buchan et al.).Our conclusions are summarized in two tables, which modify where necessary, but retain the original scheme where we consider it will serve beneficially. We also include, in anticipation, the scheme from another paper submitted in 1976 and accepted for publication as a Skrifter of the Norsk Polarinstitutt. This has already (informally) met with similar criticism, so we anticipate and attempt to justify what we have done. However, all in all, these are matters of convention rather than positive substance, and so we do not overestimate their importance. We are primarily concerned at the confusion that has arisen through the presentation of alternative schemes.In our papers we have followed current guides to stratigraphic procedure (e.g. Harland et al. 1972; Hedburg, 1976) in regarding the formation as the primary local rock unit that should be internally (relatively) homogenous, externally distinguishable and presumed to have originally been physically continuous. Other visible variations within the formation are recognized by member status. We do not regard age as a valid criterion with which to divide or combine otherwise distinctive units. We prefer to define the units by all characters (including fossils) and then begin or continue discussion on age, environment, etc. Thus a subsequent change of opinion as to the age of a particular unit does not automatically require a change in the name or rank of that unit. Indeed, after careful consideration, we decided to introduce new formational names for different, widely scattered areas on the basis of recognizable differences. A major factor in the argument of Edwards et al. is that because these formations are all about the same age they should have the same name -i.e. one should therefore extend the Spitsbergen nomenclature to the other islands of the archipelago. They argue also that it will be easier to remember fewer names, but for this function we consider that as there are already many formational names in Svalbard the memory need not be burdened with other than group names. So we unite our different formations in...