''Lithosphere'' is a mechanical concept implying strength and relative permanence. Unfortunately, the term has also been applied to the surface thermal boundary layer (TBL) and a shallow enriched geochemical reservoir, features having nothing to do with strengt. The ''strong'' lithosphere is about one half the thickness of the TBL. The bottom of the elastic plate, the maximum depths of midplate and fracture zone earthquakes, and the top of the low-velocity zone all occur at depths corresponding to the 550¿--750¿C temperature range. This may also correspond to the base of the coherent plate. Earthquakes in subducted slabs are bounded by isotherms in this temperature interval. Mantle hotter than ~650¿C cannot support long-term stresses and does not qualify as lithosphere. There is a very little ancient lithosphere (mantle colder than colder than 650¿100¿C for long periods of time), and this is not a suitable reservoir for continental flood basalts (CFB). A chemical characteristic, that is, ''enriched,'' has been attributed to old lithosphere to distinguish it from the ''depleted'' upper mantle. The continental lithosphere (CL) is often treated as a viable reservoir for CFB or, when delaminated, for ocean island basalts and enriched mid-ocean ridge basalts (MORB).

The CFB reservoir is more likely to be a hot, weak subltihospheric layer which may include the lower part of the TBL. The term ''perisphere'' has been introduced to accommodate the need for a term for a global, shallow, enriched resevoir or boundary layer. It is physically isolated from the depleted mantle (usually called the ''convecting mantle,'' ''asthenosphere'' or ''upper mantle'') not by its strength but by its weakness and buoyancy; it is chemically isolated by its location relative to subduction recycling. It has the chemical characteristics often attributed to continental lithospheric (or plume heads), but its is a permanent part of the sublithospheric shallow mantle and is constantly refreshed by recycling. It is an open system and can also be called the ''active layer'' or ''mixing zone.'' It is proposed that the depleted reservoir (MORB source or depleted mantle) is below and protected from contamination (chemical isolation) by the filtering action of this boundary layer. Depleted MORB is most prominent as fast spreading ridges, which induce or localize deep, broad upwellings.

Melts from enriched mantle (EM) are most evident at a new or slowly rifting regions, infant subduction zones, new backarc basins, slab windows, and midplate environments away from spreading induced upwelling. EM is therefore probably shallow. It is not known if volatiles and large-ion lithophiles can can recycle much deeper than ~200 km, or into the lower mantle, as is implied by some plume theories. The base of the (strong) lithosphere and plate may correspond to a phase change

If so,the correspondence among the brittle-ductile boundary (the maximum depth of earthquakes), the top of the low-velocity zone, and the elastic plate thickness can be understood. Candidate phase changes include dehydration and clinoenstatite to orthoenstatite since these occur near 600¿C. Another rheological boundary may set in near the solidus, but silicates lose most of their strength at absolute temperature about one half the dry solidus temperature.

The region of the subcontinental mantle between ~600¿C and the solidus may provide some of the material in continental magmas, but this cannot be considered part of the the continental plate. The ''continental lithosphere'' reservoir of petrologists is actually a weak enriched layer that may spread across the top of the convecting mantle. This is the perisphere. Its existence makes it possible to understand CFB and ocean island chemistry and kinematics without postulating plumes from the lower mantle, plume heads, fossil plum heads, or delaminated CL. The upper mantle is inhomogenous in chemistry. ¿ American Geophysical Union 1995