I’ve been told that one lucky astronaut, during an HST service mission, had to have his head and eyes right in the focal plane where a CCD goes. For a time, he was literally looking through the Hubbell into outer space. I give this story a better than 50% chance of being apocryphal, but it’s neat to think about nonetheless.
Well, the Refractive index of air is about 1.00029, so it doesn’t slow light down by much. For example, a photon would take 1000 years and 106 days to pass through a 1000 light year wide region of air.
But there’s a problem with having large regions of space that even approach the density of air: gravity.
A spherical mass of air about the size of the orbit of Jupiter (2600 light seconds) has a Schwarzschild radius of about 1300 light seconds. In other words, it’s a freaking black hole.
Astronomically big, atmospherically dense regions of gas are not stable. If they don’t turn into black holes, you’ll get stars forming from them for sure.
Because parallax gives you distances relative to the radius of the Earth’s orbit. Astronomers started using parallax to measure distances before this quantity was known, so they couldn’t translate these measurements to other, more common units of length.
Of course that’s not an issue now, I imagine the radius of the Earth is known down to a few meters. But people got used to quoting distances in parsecs, so they continue to do so.
I know how they do it. I don’t know how distance that great are given in the astronomical literature. I read popularizations, which use light years (though explain what is meant by parsecs) but I don’t read journal articles.
I expect that there are distances (like in the local group where gravity overwhelms expansion) where redshifts are irrelevant and which are too far for parallax.
Yes, longer distances within the Galaxy are generally measured in kpc (kiloparsec), and distances to nearby galaxies are generally measured in Mpc (megaparsec).
My guess is, it’s more straightforward to express error bars for the parallax than for distance.
Not even that good. Parallaxes from Earth-based telescopes are good to around 100 parsecs. Further out from that, the error bars become too large. Hipparcos increased that to 300-400 parsecs, but even there there’s a bit of argument about some of its results, especially the distance it found for the Pleiades.
Now the Gaia mission should get us to that 1000-parsec distance and possibly further. But it’s not going to launch for at least 3 years and, if it runs true to form for these kinds of things, data won’t be available for several years after that.
I think parsecs are popular as a unit because, as Siplicio says, there was a time when the accuracy of parallax measured distances in parsecs was better than the accuracy of the conversion between parsecs and other distance measurements, and because there still has never appeared another unit that is very advantageous relative to parsecs. IMHO they should use SI, like almost all other fields of science. The only three fields of science I know of that cling to non-SI units are astronomy, aerosol science which uses the cgs system, and spectroscopy which uses inverse centimeters and prints scales increasing to the left.
To actually translate parallax measurements into physical distances, you need to know the absolute value of the Astronomical Unit (better known as the radius of Earth’s orbit.) Currently the best measurements of this quantity are from radar-ranging of Venus; we send a radar pulse out to Venus and figure out how long it takes to get back. Multiplying this time by the speed of light gives us the measure of the AU. In some sense, all current distance measurements depend on knowing the speed of light, since knowing the value of the AU is the first step on the cosmic distance ladder. However, precise measurements of transits of Venus allow for measurements of the AU without knowledge of the speed of light, so we wouldn’t be entirely lost without knowing what c is.
The error bars are, indeed, much more straightforward for the angles than for the distances. To convert a parallax into a distance, you need to take the tangent of an angle which is extremely close to 90 degrees, which means that errors are magnified to an insane degree. If I measure a parallax as being something like 0.001 ± 0.001 arcseconds, then I literally have infinite error bars on the distance measurement. Even for measurements better than that, Gaussian error profiles on the parallax will lead to highly non-Gaussian error profiles on the distance, which significantly decreases the value of error bars, since most of the techniques for analyzing them assume Gaussians.
Parallax angles are so small that astronomers don’t bother with the tangent function. They just take the inverse of the parallax as the number of parsecs. You can see this in footnote 7 on the page I linked to above where they say that to get the number of lightyears, just divide the parallax into 3.26.
Don’t know how they handle converting error bars to parsecs. I suspect they don’t do it at all.
There’s recently been a very involved new analysis of old Hipparcos data for steller distances, which has created a sensation because the estimates of a number of pretty well known distances have shifted surprisingly. I think the Pleades is the best known target to have undergone a big shift. The analysis is a computer reduction of many, many observations by a regression method that considers all the data taken together in the estimation of any one of the distances. The error analysis component of this project is probably worth a dozen theses. Hipparcos is a satellite specifically for surveying the sky by parallax measurement.
From the Wikipedia article about Hipparcos:
"There were questions over whether Hipparcos has a systematic error of about 1 milliarcsec in at least some parts of the sky. The value determined by Hipparcos for the distance to the Pleiades is about 10% less than the value obtained by some other methods. As of 2007, the controversy remained unresolved.[4][5]
In order to resolve these issues, Floor van Leeuwen published reevaluated results.[6][7]"