Our analysis indicates that this is one of the rare detached systems in this spectral range, with the components still close to the ZAMS: the evolutionary models by Claret and Giménez (1992) yield an age of 2.5 yrs, while new models by Claret (1995) indicate an even younger system at an age of 1.6 yrs. Both sets of models were calculated with the chemical composition (X=0.70, Z=0.02), and with core overshooting.
Amongst the well known (errors in the absolute dimensions 2%) systems with both components having M , V 3903 Sgr is about the youngest, the one with the most massive primary and where the masses of the components are most different (important in studies of evolutionary tracks). The other systems are EM Car, , , 4.5-5 yrs (Andersen & Clausen 1989) and Y Cyg, , , 2-3 yrs (Simon et al. 1994, Hill & Holmgren 1995). DH Cep (Hilditch et al. 1996) is a non-eclipsing ellipsoidal variable at approximately the same age as V 3903 Sgr and with larger masses (, ) but with uncertainties 5% due almost entirely to the determination of the orbital inclination (at i = , a error results in a 5% change in the masses).
The calibration for effective temperatures and absolute magnitudes for 30 000 K is problematic, due to the lack of reliable empirical determinations in this interval. Schönberner & Harmanec (1995) present a calibration for 10 000 K 30 000 K. In their study they included preliminary absolute dimensions of V 3903 Sgr (Vaz et al. 1993), but the results were not included in their calibration because they were not very consistent. The intrinsic values of - , (=5 040/ and , obtained using Tables 9 and 10 and E(b -y)=0.32 (Sect. 4), are presented in Table 12. The definitive absolute dimensions determined in the present work place both components so nicely in the diagrams presented by Schönberner & Harmanec (1995) that we dare suggesting Table 12 as an addition for their Table 2 (which should replace the corresponding entries in the Table 3 of Philip & Egret 1980), extending the range of effective temperatures of their calibration up to 38000 K. V 3903 Sgr is one of the least evolved systems at this mass range and we believe that the present solution is rather robust.
Table 12. Calibrated , and absolute visual magnitude for the components of V 3903 Sgr.
The preliminary absolute dimensions of V 3903 Sgr (Vaz et al. 1993) were also used by Hilditch et al. (1996), and the inconsistency detected by Schönberner & Harmanec was again noted. However, the definitive solutions presented here are fully consistent, also when analysed with the evolutionary models of Schaller et al. (1992), used by Hilditch et al. (1996) to interpret DH Cep, which somewhat confirms an age of 1.6 yrs for V 3903 Sgr.
The circular orbit and the synchronized intrinsic rotation velocities can be explained theoretically, despite the short age of the system. For both models by Claret & Giménez (1992) and Claret (1995) it was possible to find an isochrone for the system and individual evolutionary tracks that matched very well the results for both components. No sign of mass transfer could be detected through period changes and the absolute parameters for the non-interacting components of V 3903 Sgr should then be representative for single stars of the same mass.
V 3903 Sgr is important for the study of stellar evolutionary models, due to its very massive (but having different masses) and young components. Besides, being a member of the R association Simeis 188 (Herbst et al. 1982), the illuminating star of the bright nebula IC 4685 (Hirshfeld & Sinnot 1982, 1985), and very possibly a physical member of the Lagoon Nebula Complex (Messier 8 and NGC 6530), V 3903 Sgr is an important object for the study of this complex structure, also, where star formation still is going on (Stahler 1985).
© European Southern Observatory (ESO) 1997
Online publication: April 6, 1998