The beamsplitter is really a visual wavelength blocking filter:
The beam splitter used in the Michelson interferometer is purchased from SurplusShed.com. It is their item number L14550. The vendor does not list much information about the beam splitter. They do say the following:
"These are beauties. Front surface has a gold coating. Light reflection appears gold but if you look thru it it the image appears blue. Looks like about 50 percent reflection and 50 percent transmission... We don't have any more specifics for this... One of our customers supplied these thoughts: "I think you'll find that the
beamsplitters are actually dichroic beamsplitters. They function as a combination of high-pass filter and lowpass filter." Thanks for the suggestion, Rick."
As you will see later, the 50% transmission/reflection is a little iffy but it does look like Rick was right, the element is most likely a filter.
Because the Surplus Shed web page suggested the beam splitter is really some kind of optical filter I was curious what the spectra of light passing through the element looked like. To find out Dr. John Carini of the Indiana University Physics Department loaned me an Ocean Optics Spectrometer. Using it I was able to determine the bandpass of the beam splitter. It appears what Surplus Shed advertises as a beam splitter is indeed an optical filter. The transmission response of the beam splitter is shown in the following graph.
To obtain the response of the filter I used a 12 volt GE "High Intensity" incandescent bulb like used in many desk lamps. I placed the bulb approximately 4 inches away from the Spectrometer's sensor. I recorded the response of the lamp with nothing between it and the sensor. This response is the red curve in the graph. I then placed the beam splitter between the lamp and the the sensor and recorded the response. That response is shown in blue. Because the lamp does not have a constant amplitude we must normalize our data to obtain the actual response of the beam splitter. This is done by taking the ratio of the beam splitter response (Blue curve)/raw lamp response (red curve). This ratio gives us the transmission response of the filter. It is shown in Dark Green on the graph (It may appear black on your monitor).
The long wavelength cutoff frequency is around 850 nm (in the near infrared region). The short wavelength cutoff is around 450 nm (in the violet). In between these two wavelengths the transmission response is approximately 30 to 40 percent of the incident light. So the filter blocks about 60% to 70% of the visible light that hits it. Assuming this is a dichroic filter the losses are small, on the order of perhaps 2% to 5%. We'll assume a near perfect filter and estimate the reflected light by taking one minus the transmission response. So we have a visual band blocking filter with about 40 percent transmission and 60 percent refection.
So why do we see yellow for light reflected off the filter and blue for light transmitted through the filter? Looking carefully at the graph, the transmission peaks at approximately 39 percent from 550 nm to 590 nm. This is in the yellow region of the spectrum. Below 550 nm the transmission drops to around 32 to 35 percent. Above 590 nm the transmission drops even more. From 700 nm to 850 nm the transmission drops as low as 30 percent. So yellow is transmitted more than the blue or red ends. And the red end is attenuated more than the blue. So the transmitted light looks yellow because it has the highest transmission whilst the reflected light looks blue because the red and yellow are attenuated most on reflection, leaving the blue as the strongest component of the reflected spectra.
If I were to describe the beam splitter filtering properties I would call it a 35% transmission, 65% reflection visual band filter.
Bottom line: However you want to describe it, the beam splitter works very well for our needs.