The Spectroscopy of Filters
     I have always enjoyed spectroscopy in Chemistry, and have wanted to know more about the filters used in Astronomy.  In Chemistry UV/Visible spectroscopy is used to determine the concentration of materials in various solutions without having to isolate them.  A general procedure would go something like this.  Dissolve your sample up in a solvent and place it in a glass cuvette (square testtube) of known optical path, usually 1 cm.  Take a 2nd matching cuvette and fill it with the same solvent that your sample is in. This is your "standard". Using various methods, take a reading of each cuvette at each wavelength of interest.  By dividing the quantity of transmitted light of your sample by the transmitted light of your standard, the absorption of the sample is measured.  The "standard" is considered to be 100% transmission, and the absorption curve is displayed as a percentage or fraction of that. 
     Of course, the transmission of the standard is not really 100%.  The glass absorbs light, and the solvent absorbs light.  The purpose of using the standard is to eliminate those variables and isolate the amount of absorption by the sample.

In order to learn more about astronomical filters, I purchased a UV/Visible/IR spectrometer capable of providing spectra in the 300-1000nm range.  In general, the light source I use has no UV so my normal spectra are from 400-1000nm, but I have a light source capable of delivering radiation in the 300-400nm range should I need it. When I started taking spectra, something became immediately apparent from the experiment I describe above.  In the case of an astronomical filter, what is the appropriate standard?
  At first I gave it very little thought.  Since I wanted to know how the filter worked in my telescope, I chose air (essentially nothing) as my standard.  Before I put the filter in my telescope, air is what is there.  I would like to know what changes the filter makes in the optical path and so measuring the total changes appear to be the correct way to approach this. I quickly found out that the spectra I took did not quite match the manufacturers samples.  Orion's spectrum for their ultrablock filter is below left (400-700nm); my spectrum (400-1000nm) is on the right. They are quite similar in the 400-700nm (visible range) with the exception that I am reading a maximum transmission of 0.65 (65%) vs their 90%. 

While I am using an Orion filter here, I am not trying to single them out.  This behavior proved to be pretty generally true.  I was seeing 10-25% lower transmissions on all of the filters I measured.   Clearly I was encountering some sort of systematic problem .  In fact, none of the 40 or 50 samples I have taken ever measure above 80%-85%. 

      How can this be?  Well I can only see 2 choices.  My new spectrometer doesn't work right, or the folks doing these spectra are using a different standard than air for filter spectroscopy.  I realized at this point one could argue, just as in the Chemistry description above, that one should separate the effects of the filter itself from the glass that holds the filter.  After all, from Dave Hall's and my earlier study of eyepiece transmission we knew that the more air-glass interfaces there are, the more absorption of light there will be.
     At this point I decided I should contact the manufacturers and ask them to tell me how they take their spectra.  I contacted Orion on May 21, 2012 requesting information on what standard they use when taking these spectra.  On May 25, I contacted Baader, Celestron, Hands On Optics, GSO, Lumicon, and Sirius Optics requesting the same information.  Meade does not accept email requests, and I have not yet bothered to write them a letter.  Thus far I have not heard back from any of them.

     While waiting for replies, I decided to try an experiment.  What would happen if I tried some simple uncoated clear glass as my standard.  A microscope slide came to mind and I bought some.  Since they are uncoated, one would expect them to overestimate the correction with respect to a coated piece of glass. The purpose of the coating is to minimize light losses. Here are the results for the two filters above. 

Sure enough, my spectrometer can show above 80% (0.8) transmission.  My spectra now look very much like the manufacturer curves, perhaps the correction is a little bit large as they look to be slightly higher than the manufacturer curves.  The Orion ultrablock shows I can still get down to 0% transmission also.  This leaves us with a question.  Which is the "correct" standard?  In my mind, using a coated piece of glass as a standard would be correct if I constantly put one of those on the end of every eyepiece.  Thus when I removed it and put a filter in its place  these curves would accurately represent the reduction in the light transmission of the system.  Personally I don't do that.  In fact, I don't know where to buy such a coated optical dust cover for my lenses or I would buy one to use as a standard, perhaps they are available somewhere.
    On my telescopes, there is nothing but air at the end of my eyepieces and the stronger absorptions seen with my air standard ought to represent that case well.   Unfortunately that means that the light losses for using a filter are significantly larger that the manufacturer curves seem to indicate.  Of course all of this is subject to modification when the manufacturers tell me how they run their spectrometers and what standards they use for setting 100% transmission.  When they tell me, I will post it here (see below).

Direct comparison of the above data for the Mars 2003 filter.  In this comparison, the spectra of the manufacturer, the air standard and glass standard are superimposed on the same graph.  Since the manufacturer gives no IR data, the spectra were stopped at the point the manufacturer data ended.
The grey is the manufacturer spectrum.  The yellow is my spectrum when using glass as a standard.  The red is when air is used as the standard.  This implies that in your telescope where you are replacing air with this filter, your transmissions will resemble the air standard most closely. 
Manufacturer Responses:
June 2:

Hands On Optics:

Dear Dr. Duchek,

We are factory representatives for several manufacturers.  The statistics

that we publish are generated by the manufacturers.

We do no spectral testing here.

Gary Hand

  I then requested that the contact their manufacturer with my request.

Celestron responded to tell me that they have referred my request to their R&D group.

June 16:

Dear Mr. Duchek,

Thank you for your email to Orion.  The transmission graphs are provided by the supplier.
I do not have specific information as to the method used.  I regret I am not able to provide
more information at this time.

Clear Skies,

Rich R.
Orion Technical Support,, (800)-676-1343

07/02/12 -Since I wrote this article I have talked to some knowledgeable amateurs who agree me about air being the logical and reasonable standard.  I have also run across a couple of filters that do match the manufacturer standards given in that a certain wavelength be 95% transmission.  On some filters where I have 3 different "identical" filters, I find they vary in this respect.   I have revised the database to only included the air standard, and will let the transmissions fall where they may.  (The spectroscope will be calibrated to an air standard before any spectra are taken.)