It was presented in the work of Tomov & Tomova (1992) that the orbital variations of the radial velocity of the HeII 4686 broad component are opposite in phase with those ones of the giant and are determined by the motion of the hot companion. The radial velocity variations of all nebular lines (the lines of ionized metals are an exception) have practically the same phase shift in relation to the radial velocity variations of the two stellar components (Fig. 3 (click here)). This means the radial velocities of all these lines are determined by the same flow in the circumstellar nebula which, as a whole, is not at rest and there is probably a pronounced large-scale motion in it. The phase shift of the radial velocity curves shows the gas flow is in correlation with the orbital motion. However, the different amplitudes probably indicate some stratification in the direction of the motion causing the various lines to be appear in areas with different velocities.
The described behaviour of the radial velocities during the orbital cycle is related to a binary system, where both components have a stellar wind. Then the two winds must collide head on, as presented in a simplified approximation by Girard & Willson (1987). A nebular region with a high density and a low velocity of the gas motion is formed as a result of that collision. That is why it was supposed (Tomov 1993b) that the radial velocities of all of the visual narrow lines of AG Peg are probably determined by the motion in the conical part of the nebular region of this system.
The analysis of the emission line spectrum (Sect. 3) showed that
the intensity of all lines has decreased. The possible excitation
mechanisms in the nebular region include
shocks and photoionization. That is why the total flux of the emission
lines appearing in this region will consist of two parts:
one determined by radiative heating
and the other - by shock excitation. The interpretation of the decrease of
these fluxes requires modelling in detail of the emitting region.
This is a problem of a theoretical nature, including computation
of its shape and size as well as the velocity and density distribution.
However, it is possible to show by means of elementary calculation that
the fluxes would be decreased by a factor close to the mean observed
decrease factor of all elements if ionization is realized only by shocks.
The mass-loss rate of the giant and the velocity of its wind are considered
to be constant having values 210-7 
yr-1 and 20
kms-1 (Vogel & Nussbaumer 1994;
Mürset et al. 1995). The wind parameters of the hot companion
are presented in the last section. Then the sum of kinetic energies of the
winds in the two moments considered by us is calculated to be
4.911034 ergs-1 and 2.771034 ergs-1
respectively. These are upper limits, since the wind velocity
components normal to the nebular region surface in its predominant part are less
than the velocisties themselves. The ratio of these amounts is 1.8 and
the average of the observed values of the decrease factor of
the different elements is 2.5.