Quote:
Originally Posted by rtomes
Quote:
|
Originally Posted by Nereid
(my bold)
It seems my question was not sufficiently clear; let me try again, with specific reference to galaxy redshifts.
Would you please walk through the steps involved in making an estimate of a galaxy's redshift, indicating in which steps the major sources of uncertainty (in the galaxy redshift) arise.
For those key steps - in terms of the source of uncertainty in an estimate of a galaxy's redshift - please indicate how that uncertainty is estimated.
For a set of 'galaxy redshift' inputs, each derived redshift having an estimated uncertainty, please describe the steps involved in estimating any periods in the set, and indicate in which steps the major sources of uncertainty (in the estimated period) arise.
For those key steps - in terms of the source of uncertainty in an estimate of a redshift period(s) of the set - please indicate how that uncertainty is estimated.
|
You get a telescope and point it at the object, and it has a slit and a prism or other device for separating the spectrum into colours. You put a device that can measure the intensity into the spectrum after focusing it, and take readings all the way along. These days I think that CCD devices are used to pick up the whole spectum at once. If it is a CCD device then the number of pixels is obviously critical to the resolution. You need to have calibrated the spectrometer with some ordinary spectrum on Earth first. To determine the redshift you need to do a correlation of some similar reference spectra that has lines expected to be found (e.g. Hydrogen, Helium, metals etc) by sliding the two until agreement is reached (this will be done mathematically today I imagine - that is most easily achieved by converting the spectrum to a log basis). The amount of slide is the redshift as measured - the difference between the wavelengths as a fraction of the wavelengths. You then adjust for the part of the vector of Earth's motion that is toward or away from the object to get a solar system based redshift.
The biggest problems in getting a redshift are I imagine the light gathering power and the resolution of the device for measuring the light (CCD?). It may need a long exposure if it is a distant object. If it is a dim object you cannot use a wide dispersion or there will be too little light captured. If you measure the wavelength of the light to an accuracy of 1A then you can calculate the spectrum to about a z of +/-.0002 from a single line. If you use many lines and there are no systematic errors (everything has been well calibrated) then this might be improved a bit by averaging many lines. Of course spectral lines are not infinitely small, as for a galaxy there might be 500 km/s dispersion which is about 10 A in the spectral lines. This is less of a problem for a star obviously. So you aim to get the centre of the peaks or dips depending whether it is emission or absorption. That will be the point of maximum correlation.
The best way to measure the uncertainty is by taking multiple measurements of a few different objects with different equipment. Then you can get a s.d. of the errors. This is more certain that estimating from the components effects. You will not expect to get more accurate than the resolution of the measurements permits.
[snip] |
How closely does this correspond to what Tifft and Croasdale (or rather the astronomers who actually took the relevant observations) did, to obtain estimates of galaxy redshifts?
How closely does this correspond to what Croasdale reports, in the paper you cited, are the greatest sources of uncertainty?
What does Croasdale's paper say about how he measured, and addressed, uncertainty, at the level of individual galaxy redshifts, and as a set?