[RadioAstronomer]

The only possible criticism is that the error bars seem to be too large. To my eye, the chi-squared per degree of freedom seems considerably less than one.

I'll have to think about that for a bit.

[Physicist]

I'll have to think about that for a bit.

I'll try to dig up the paper itself; I'm sure it will say what the chi-squared actually is.

I base my estimate on the expectation that, if the errors are correctly estimated, somewhat more than a third of the data points will be more than one sigma away from the curve. In the above plot, only three or four data points are so aberrant out of 20.