7. Concluding remarks
The magnitudes of R 84 reported in the literature are based on classical photometry using large apertures of some 15 in diameter. However, this work has shown that about 30 stars populate these apertures and the derived magnitudes are overestimates although R 84 dominates the cluster. In deriving the magnitudes of R 84 we have surpassed this crowding problem. On the other hand, if the late-type star is confirmed to be a line-of-sight companion, this means that the visual magnitudes so far given for R 84, including our relatively lower values (Table 2), are all upper limits. Assuming an apparent magnitude of V = 13.2 for the M2 supergiant (see Sect. 6.2) leads to a V = 12.6 for R 84, based on a global magnitude of 12.10 (Table 2) for both stars. This corresponds to an absolute magnitude of = -6.6, which is fainter than the recently derived value of -7.0 (Crowther et al. 1995 ).
A remarkable point is the apparent absence of O type stars in the OB association LH 39. Schild (1987 ) derived the spectral types of the 16 brightest blue stars of the association. In this sample 13 stars are B types (only 3 of them main sequence), there are 2 A supergiant stars, and R 84. Stars #35 and #36 that we have observed towards the core of the association are also B types (Sect. 4). One of them may be a double system with spectroscopically identical components. Several other stars in the association, for which no spectroscopy is available, have a reddening-free Q index consistent with a B type classification (Sect. 6.3). For instance, stars #3 and #14 have V = 16.24 and 15.83 (Table 2) corresponding to absolute visual magnitudes of about -3.0 and -3.5 respectively. These are too faint for O types and would rather indicate main sequence B stars.
If there were any undetected O stars in this association, they should be fainter than the main sequence B types, and this is difficult to admit since the reddening is pretty uniform over the association and we do not expect important local extinction in this rather evolved association. Therefore, we conclude that the turn-off of this association lies around B0. This entails a rather provocative question as to the progenitor of R 84. Could R 84 come from a B type star? Of course one is inclined to admit that the progenitor of R 84 was the most massive star of the cluster, a supposedly O type. Generally speaking, the formation of O type stars is a collective process, as indicated by recent observational and theoretical works (see Heydari-Malayeri 1996 and references therein). But there is no observational evidence supporting the existence of O stars in LH 39, as is the case in other LMC OB associations (e.g. LH 117 and LH 118, Massey et al. 1989 ).
According to the Z = 0.008 evolutionary tracks of Schaerer et al. (1993 ), the mass limit capable of forming a W-R star is slightly above 40 and R 84 lies very close to the 40 track. Although, very massive Galactic late-type WN stars are probably still in the H-burning stage (Rauw et al. 1996 ), it seems unlikely that less massive stars in the LMC will reach the W-R stage prior to the He-burning phase. Therefore, the total H + He-burning lifetimes for a 40 star will set an upper limit of 5.3 Myr to the age of R 84, whereas most of the association members are consistent with an age of about 12 Myr. Unless the mass limit to form a W-R star is considerably lower than 40 , we are forced to admit that the stars in LH 39 are not coeval.
The presence of a low mass M supergiant in this OB association is another challenging problem since Humphreys et al. (1984 ) showed that red supergiants and W-R stars were anticorrelated in the OB associations of M 33 (metallicity similar to the LMC). Its formation might be more easily understandable if it belongs to a binary system where the physical effects of mass exchange would considerably alter the evolutionary paths. If the partner is effectively the Ofpe/WN9 star, such a scenario could also provide an explanation for the present status of R 84.
© European Southern Observatory (ESO) 1997
Online publication: April 20, 1998