5.5.4. Why is there lingering uncertainty over D?

Today it is widely agreed that D is low towards a few quasars. There remains uncertainty over whether there are also cases of high D, for the following reasons:

• measurements have been made in few places;

• contamination of D by H looks very similar to D, and resembles high D;

• both the low Y_p values reported during the last 25 years, and the ⁷Li abundance in Spite plateau halo stars, with no correction for depletion, imply low _b , low , and high D/H for SBBN; and

• the first claims were for high D.

In most cases, the apparent conflicts over D/H values concern whether the absorption near the expected position of D is mostly D or mostly H. Steigman [109] and all observational papers discussed this contamination of D by H.

Carswell et al. [60] noted that contamination was likely in Q0014+813 and hence the D/H could be well below the upper limit. Songalia et al. [71] stated: ``because in any single instance we can not rule out the possibility of a chance H contamination at exactly the D offset, this result [the high D/H] should be considered as an upper limit until further observations of other systems are made.'' Burles et al. [72] showed that Q0014+813 is strongly contaminated, does not give a useful D/H limit. For Q1718+4807 we [68] and Levshokov, Kegel & Takahara [100] have argued that contamination is again likely.

There are many reasons why contamination is extremely common:

• H absorption looks just like that from D,

• H is 30,000 times more common,

• spectra of about 50 quasars are needed to find one example of relatively uncontaminated D,

• high signal to noise spectra are needed to determine if we are seeing H or D, and

• these spectra should cover all of the Lyman series and metal lines, because we need all possible information.

It is essential to distinguish between upper limits and measurements. There are only two measurements (Q1937-1009 and Q1009+2956). They are measurements because we were able to show that the D absorption line has the expected width for D. All other cases are upper limits, and there is no observational reason why the D/H should be at the value of the limit. In many cases, all of the D can be H, and hence and D/H = 0 is an equally good conclusion from the data.

Only about 2% of QSOs at z 3 have one absorption systems simple enough to show D. All the rest give no useful information on D/H. Typically, they do not have enough H to show D, or there is no flux left at the position of D. In such cases the spectra are consistent with high, or very high, D/H, but it is incorrect to conclude that D/H could be high in 98% of abosrption systems because these systems are not suitable to rule out high D/H. Rather, we should concentrate on the few systems which could rule out both high and low D/H.

We will continue to find cases like Q1718+4807 which are consistent with both low and high D/H. As we examine more QSOs we will find some cases of contamination which look exactly like D, even in the best spectra, by chance. But by that time we will have enough data to understand the statistics of contamination. We will know the distribution function of the contaminating columns and velocities, which we do not know today because the D/H absorbers are a rare and special subset of all Lyman limit absorbers. When absorbers are contaminated we will find a different D/H in each case, because the N_HI , velocity and width of the contaminating H are random variables. But we will be able to predict the frequency of seeing each type of contamination. If there is a single primordial D/H then we should find many quasars which all show this value, with a tail of others showing apparently more D/H, because of contamination. We will be able to predict this tail, or alternatively, to correct individual D/H for the likely level of contamination. When we attempted to correct for contamination in the past [53], [110], [68], we used the statistics of H I in the Ly because we do not have equivalent data about the H I near to the special LLS which are simple enough to show D/H. Such data will accumulate at about the same rate as do measurements of D/H, since we can look for fake D which is shifted to the red (not blue) side of the H I.

There are large differences in the reliability and credibility of different claimed measurements of D/H in quasar spectra, and hence much is missed if all measurements are treated equally. It also takes time for the community to criticize and absorb the new results. Early claims of high D/H [103], [111] in Q0014+8118 are still cited in a few recent papers, after later measurements [72] with better data, have shown that this quasar gives no useful information, and that the high D/H came from a ``spike'' in the data which was unfortunately an artifact of the data reduction.

In summary, the lack of high quality spectra, which complicates assessment of contamination by H, is the main reasons why there remains uncertainty over whether some absorbers contain high D.