Identifying absorption lines with confidence; instrument response curves revisited

Here's a profile I produced of the hot young star Sirius:

As you know, the "dips" in the graph show where specific wavelengths of light were absorbed by the gas in the upper atmosphere of the star, and these can tell us exactly what the gas is made of. In this case, the dips indicated by the white triangles show 3 of the main lines for Hydrogen. This is what you would expect because a young star is mostly made of hydrogen.

But what about the 2 dips by the green triangles? 

Part of the problem of identifying these is that this graph is uncalibrated; the curves show what the camera sensor (Nikon D60) is sensitive to but this is not linear across the wavelengths. As we saw before (previous post) a camera CCD is not sensitive to lower and higher wavelengths, and also not uniformly sensitive even in the normal visual range.  One way to find out what is real and what is not is to compare the profile with a reference.  Here's the Library Profile of a similar star on RSec:

You can see the problem; they look totally different. The answer is to snip out the high and low wavelengths so that it matches the original profile. And then subtract the library profile from the raw. (Instrument Response Calibration; previous post) 

Here's my profile after adjustment:

This is now a true representation of the various wavelengths. 

Now let's add back in the snipped library profile:

The reference profile (blue) confirms the strong hydrogen lines and also show that the dip by the left hand green triangle is just noise.  The right hand green triangle dip is simply the "remnant" of the huge dip in the raw profile, and is not associated with any real feature.

The upshot is, I won't be making any more spurious claims as to the chemical composition of the stars, because I'll be able carefully check the results against professional spectra. Marvellous!