The emission measure Ne2 dh, which is so useful in reducing XUV as well as radio data, is redefined as f(T) dT where f(T) is called ‘thermal emission measure’. Theoretical predictions for f(T) on the basis of a simple, one dimensional, steadily expanding atmosphere are presented. Depending upon the boundary conditions, and essentially upon the mass flux, two very different behaviours show up.

1. (1) With a mass flux compatible with an extrapolation of the solar wind flux, f(T) would correspond to a transition region controlled by a constant conductive flux. It can be fitted, as was shown e.g. by Athay (1966), with the XUV observations of the integrated disk for lines formed at temperatures higher than about 2 × 105 K.

2. (2) But only with a mass flux enhanced by a factor of 50 to 100, can we interpret the radio spectrum and the XUV observations of lines formed at temperatures less than 2 × 105 K.

The suggestion is made that a single model can reconcile both behaviours: below 2 × 105 K the emitting region is squeezed so that it covers 1–2% of the surface that it occupies at higher temperatures where all structures begin to merge, filling the whole corona. This variation with height compares with usual models derived from spatially resolved observations; as usual the underlying magnetic field is expected to support this channelling.