In order to have sufficient UV flux to reach adequate signal-to-noise in the short times available when observing from low orbit (typically 30 minutes), only relatively bright stars (V brighter than 8) of spectral types A0 and earlier and E(B-V) less than about 1.5 were considered. From a database of objects meeting these criteria the pre-mission science timeline was crafted to 1) sample the galactic plane as uniformly as practicable, 2) explore sight lines of diverse chemical composition and morphology, and 3) measure the shape of the UV polarization spectrum through the full range of known lambda_max.
An attempt was made to avoid stars with known intrinsic polarization effects. However, the need for substantial UV flux which required use of early type stars made this difficult.
Additionally, the Astro instruments were coaligned, and some targets selected by the other instruments turned out to be useful for this program.
The Astro-1 mission was plagued by a host of problems, although several stars in this program were observed, but the Astro-2 observing program was accomplished to a remarkably high degree just as planned preflight.

DATA SHOWN HERE

The stars for which WUPPE obtained useful data on UV polarization which is largely of interstellar origin are listed among the 4.5 Program class, although some were originally listed in another target class and were observed as part of another WUPPE or other ASTRO-2 instrument program.
Most of the visual polarimetry is from the University of Wisconsin's Pine Bluff Observatory, PBO, using the UW spectropolarimeter, HPOL.
Two southern stars were observed with HPOL on the WIYN telescope.
Some other southern objects were observed at the Anglo-Australian Telescope, and from Brazil by Magalhães.

Both the WUPPE and the HPOL data have been processed using a software package developed for spectropolarimetry at the UW. The data reduction procedures and the error analysis are described in Nordsieck et al. (1994) for WUPPE and in Wolff, Nordsieck, and Nook (1996) for HPOL data.

SUMMARY OF ASTRO-1 and ASTRO-2 OBSERVATIONS

Astro-1 WUPPE data showed that the extrapolation of the Serkowski curve did not reliably predict the UV polarization. In particular three, perhaps four, of six objects reported in Clayton et al. (1992) had decidedly higher UV polarization than predicted by the Serkowski formula as did one other analyzed for another program, Kappa Cas (Taylor et al. 1991b). We refer to stars which show this excess as "super-Serkowski" objects. Another object primarily affected by interstellar polarization, Theta Mus (Schulte-Ladbeck et al. 1992), was reasonably well represented by the WLR Serkowski curve. [Consistent with this, observations by the HST FOS (CWAL and references therein) found that HD7252, HD204827 and perhaps BD+64^o106 also showed excessive UV polarization while HD98695 and HD161056 were matched by extrapolating the Serkowski formula.]
There was a weak tendency for the super-Serkowski objects to be those with the lower values of lambda_max.

On Astro-2 WUPPE obtained spectropolarimetry for 20 new objects and added more data for six previously observed objects (WUPPE96). [WUPPE also observed several other targets which might be useful for this program; all these were selected by the other ASTRO instruments, so the observations are generally are not as good quality as most in WUPPE96; we are obtaining visible observations of these stars to make a proper analysis.]
In the new sample nine objects are well represented by the extrapolation of the Serkowski curve, and we confirm this in another previously observed object. However, three others show a clear tendency to have excess UV polarization relative to extrapolation of the Serkowski curve, and in in another three the effect is weakly detected. One star, HD62542, which had been thought to be adequately matched by the Serkowski formula from a short Astro-1 observation, appears with improved signal-to-noise to be a super-Serkowski object. Furthermore, the Astro-2 results confirm that there is some excess polarization between 2000Å and 3000Å in the Astro-1 object HD197770 to which can be added three additional objects in the Scorpius-Ophiuchus region.
We also obtained new data, including polarimetry from HPOL, for three additional objects which have either insufficient signal-to-noise in the UV to make a reliable judgement or are too likely to be affected by intrinsic polarigenic effects to be useful as diagnostic tools in the UV.
Finally, we have included data from objects which though predominantly polarized by the interstellar medium are nonetheless somewhat affected by intrinsic effects. They are presented in this list for completeness; attempts to use them in modeling the interstellar polarization should take heed of our caveats.

DISCUSSION

CWAL suggest that the presence of the excess UV polarization is related to the value of lambda_max.
The new data show a similar effect: when polarization at 1670Å (6 microns^-1), is plotted versus the wave number of the peak polarization for all the stars with data now available, we find that for objects with lambda_max greater than about 6000Å the original Serkowski formula with width factor K=1 adequately reproduces the far UV polarization. [However, the polarizations are all small and the errors are such that this is not a particularly strong statement.]
For lambda_maxless than 5500Å the modified Serkowski formula, using one or another of the CWAL or WLR forms of K, works for some objects but not for others.
Thus there appears to be substantial, real variation in the far UV polarization relative to the visual. This may indicate that the small and large aligned grain populations are somewhat independent and so the visual and near IR polarization, as parameterized by the value of lambda_max, cannot describe the UV polarization adequately. We may have in this variation of UV polarization relative to visual a new diagnostic of varying conditions in the interstellar medium.

SMF were able to identify several regions on the sky in which either large or small lambda_maxvalues prevailed. CWAL, considering those objects for which UV data existed in 1994, suggest a low lambda_max- excess UV polarization correlation, although all of the super-Serkowski objects then known were in galactic longitude 100 - 150 deg.
With the new data in WUPPE96, we find that the concentration of stars with highest UV polarization remains in that same portion of the northern Milky Way, but in it are several objects which are adequately represented by the Serkowski extrapolation. Indeed, in the immediate vicinity of the cluster NGC1502 there are both high and low UV polarization objects all of which have roughly the same value of lambda_max, and likewise a region near longitude 260 has both high and low UV polarization objects.

See Anderson et al. AJ 112, 2726 1996