2007A&A...461..593G

We study the UV spectral behavior of the recurrent nova T Pyx during 16years of IUE observations. We examined both the IUE line-by-line images and the extracted spectra in order to understand the reality and the origin of the observed spectral variations. We compare different extraction methods and their influence on the spectrum of an extended object. The UV continuum of T Pyx has remained nearly constant in slope and intensity over this time interval, without any indication of long-term trends. The reddening determined from the UV data is E_B–V=0.25 ±0.02. The best single-curve fit to the dereddened UV continuum is a power-law distribution ∝λ^–2.33. The tail of this curve agrees well with the B, V, and J magnitudes of T Pyx, indicating that the contribution of the secondary star is negligible. One peculiar aspect of T Pyx is that most emission lines (the strongest ones being those of CIV 1550 and HeII 1640) show substantial changes both in intensity and detectability, in contrast to the near constancy of the continuum. Several individual spectra display emission features that are difficult to identify, suggesting a composite spectroscopic system. We tentatively ascribe the origin of these transient emission features either to loops and jets from the irradiated secondary or to moving knots of the surrounding nebula that are (temporarily) projected in front of the system. The inspection of all IUE line-by-line images has led to the detection of emission spikes outside the central strip of the spectrum, which in some cases seem associated to known emission features in the (main) spectrum. A comparison with other ex-novae reveals a surprising similarity to the spectrum of the very-slow nova HR Del, whose white dwarf primary has a mass that is allegedly about one half that of T Pyx.