Quote:
Originally Posted by rtomes
Well I have come across several small samples that are stated to be more accurate than 10 km/s, and it seems that H radio (21 cm) measurements can achieve accuracies of around 3 km/s. Obviously I will have to accept the best available, but it is impossible to detect a 72 km/s periodicity in a sample with 30 km/s accuracy, the minimum requirement being 18 km/s.
There are ways of processing spectral data to get greater accuracy by averaging many lines and doing interpolation, so I may have to resort to this sort of thing on some large databases, but would prefer to not have to do that.
|
What Stupendous Man was talking about is the issue that spiral galaxies, in particular, have problems of structure. The emission-line distribution is patchy in ways not directly related to the central velocity, and the dust is often preferentially located on one side of the spiral arms, giving a further bias. Similar problems affect 21 cm H I data - there are arms and blobs whose detailed distribution affects any particular measure of central value. I put
some effort into this question several years ago. The upshot is - for long-slit emission-line spectra of spirals in pairs, several plausible measures of central velocity differ among themselves, with a standard deviation of 34 km/s. The data allow measures which are
repeatable to a much tighter level, but the limit is in whether one can find a unique recipe relating the data to a unique central redshift in the presence of dynamical substructure.
(I do recall that the referee complained that I so much as cited Arp on the question of redshift asymmetries in pairs, which appear in practically everyone else's data as well. I complained back.)