5.1. ISM D/H
Observations of D in the ISM are reviewed by Lemoine et al. . The first measurement in the ISM, D/H = 1.4 ± 0.2 x 10-5 , using Lyman absorption lines observed with the Copernicus satellite , have been confirmed with superior HST spectra. A major program by Linsky et al. ,  has given a secure value for local ISM (< 20 pc) D/H = 1.6 ± 0.1 x 10-5.
Some measurements have indicated variation, and especially low D/H, in the local and more distant ISM towards a few stars , . Vidal-Madjar & Gry  concluded that the different lines of sight gave different D/H, but those early data may have been inadequate to quantify complex velocity structure . Variation is expected, but at a low level, from different amounts of stellar processing and infall of IGM gas, which leaves differing D/H if the gas is not mixed in a large volume.
Lemoine et al.  suggested variation of D/H towards G191-B2B, while Vidal-Madjar et al.  described the variation as real, however new STIS spectra do not confirm this, and give the usual D/H value. The STIS spectra  show a simpler velocity structure, and a lower flux at the D velocity, perhaps because of difficulties with the background subtraction in the GHRS spectra.
Hébrand et al.  report the possibility of low D/H < 1.6 x 10-5 towards Sirius A, B.
The only other instance of unusually low D/H from recent data is D/H = 0.74+0.19-0.13 x 10-5 (90%) towards the star Ori . We would much like to see improved data on this star, because a new instrument was used, the signal to noise is very low, and the velocity distribution of the D had to be taken from the N I line, rather than from the H I.
Possible variations in D/H in the local ISM have no obvious connections to the D/H towards quasars, where the absorbing clouds are 100 times larger, in the outer halos young of galaxies rather in the dense disk, and the influence of stars should be slight because heavy element abundances are 100 to 1000 times smaller.
Chengalur, Braun & Burton  report D/H = 3.9 ± 1.0 x 10-5 from the marginal detection of radio emission from the hyper-fine transition of D at 327 MHz (92 cm). This observation was of the ISM in the direction of the Galactic anti-center, where the molecular column density is low, so that most D should be atomic. The D/H is higher than in the local ISM, and similar to the primordial value, as expected, because there has been little stellar processing in this direction.
Deuterium has been detected in molecules in the ISM. Some of these results are considered less secure because of fractionation and in low density regions, HD is more readily destroyed by ultraviolet radiation, because its abundance is too low to provide self shielding, making HD/H2 smaller than D/H.
However, Wright et al.  deduce D/H = 1.0 ± 0.3 x 10-5 from the first detection of the 112 µm pure rotation line of HD outside the solar system, towards the dense warm molecular clouds in the Orion bar, where most D is expected to be in HD, so that D/H HD/H2. This D/H is low, but not significantly lower than in the local ISM, especially because the H2 column density was hard to measure.
Lubowich et al. ,  report D/H = 0.2 ± 1 x 10-5 from DCN in the Sgr A molecular cloud near the Galactic center, later revised to 0.3 x 10-5 (private communication 1999). This detection has two important implications. First, there must be a source of D, because all of the gas here should have been inside at least one star, leaving no detectable D. Nucleosynthesis is ruled out because this would enhance the Li and B abundances by orders of magnitude, contrary to observations. Infall of less processed gas seems likely. Second, the low D/H in the Galactic center implies that there is no major source of D, otherwise D/H could be very high. However, this is not completely secure, since we could imagine a fortuitous cancellation between creation and destruction of D.
We eagerly anticipate a dramatic improvement in the data on the ISM in the coming years. The FUSE satellite, launched in 1999, will measure the D and H Lyman lines towards thousands of stars and a few quasars, while SOFIA (2002) and FIRST (2007) will measure HD in dense molecular clouds. The new GMAT radio telescope should allow secure detection of D 82 cm emission from the outer Galaxy, while the Square Kilometer Array Interferometer would be able to image this D emission in the outer regions of nearby galaxies; regions with low metal abundances. These data should give the relationship between metal abundance and D/H, and especially determine the fluctuations of D/H at a given metal abundance which will better determine Galactic chemical evolution, and, we expect, allow an accurate prediction of primordial D/H independent of the QSO observations.