Estimating the apparent rotation using the outlined methodology can be an error prone process. This is particularly true for galaxies approaching an edge-on orientation. Consider the following example using simulated galaxy galaxy019 in the application version of the GalacticEllipseApplet. Readers are encouraged to repeat this exercise by selecting that galaxy019 from the choice list in GalacticEllipseApplet.
Recall that blue denotes the side of simulated galaxies, and that gray is used to render the top and bottom of each simulated galactic disc. Galaxy019 is correctly classified as having an apparent rotation of 83 degrees. However, if the entire visible shape were to be used in making the estimate, and assuming the difference between the "top", "side" and midplane of the galaxy could not be clearly identified, then the same galaxy might be estimated to have an apparent rotation of 79 degrees as shown in the diagram above. This constitutes an error of almost 5%. Furthermore, the galaxy has moved from well within the correct 80 to 90 degree bin into the incorrect 70 to 80 degree bin.
From this example, it is reasonable to suggest that galaxies approaching the completely edge-on orientation might potentially be under represented in a study of actual galaxy images due to error in the process used to determine inclination.
In-part, this may explain the unexpected drop in the number of galaxies in the 80 to 90 degree bin as observed in the survey. Hypothetically, suppose that all galaxies determined to be in the 75-79 degree range had been placed into the incorrect bin. In this case, 5 galaxies would have been misclassified and need to be shifted to the 80 to 90 degree bin. This would result in a total of 13 galaxies occupying the latter bin. However, this number is still insufficiently low and does not completely explain the observed dip in the distribution.
What other factors might lead to edge-on galaxies being under represented in a random sample of existing galactic catalogs?
Consider NGC 4565, the so-called Needle Galaxy found in the constellation Coma Berenices. This galaxy appears completely edge-on as viewed from earth. In this orientation, well defined spiral arms cannot be seen.
Nonetheless, it is assumed to be a spiral galaxy because it meets other criteria considered to be common to spiral galaxies. In particular, NGC 4565 has:
The edge-on orientation of NGC 4565 contains a large amount of dust which blocks light and obscures our view along the galactic plane. A similar effect from our position within the Milky Way prevents us from seeing very far in the direction of our own Galactic centre.
Although NGC 4565 is large, it is a dying galaxy producing few new bright stars. This means that it is intrinsically less bright than a similar galaxy of the same radius but undergoing active star formation.
Were NGC 4565 rotated 90 degrees about an axis along the galactic plane and perpendicular to the viewer's line of site, then it would be seen as a face-on galaxy. If this were the case, the galaxy would present the observer with a larger surface area than is seen in the actual edge-on orientation. The face-on orientation would also be relatively free of intervening NGC 4565 dust obscuring light in the edge-on orientation. As a result, NGC 4565 would be brighter in a hypothetical face-on orientation.
This suggests that edge-on galaxies might be harder to detect since they are less bright to an earth bound observer. Consequently, they might be under represented in existing catalogs.
To test the hypothesis that spiral galaxies are under represented because they are harder to detect, it would be interesting to repeat the survey conducted in the present study. The revised investigation would select all galaxies known to exist within a hypothetical shell of a given thickness and a moderate fixed radius about the Earth. The study would group galaxies in larger bins to reduce the effect of miss binning galaxies at large inclinations. Would galaxies in such a sample produce a result consistent with those observed in the present survey in which galaxies were selected "at random", or would the curve look more like that predicted by the simulation, including those galaxies approaching a 90 degree inclination?
It is reasonable to suggest that some edge-on spiral galaxies, particularly distant ones, may not be immediately identified as spiral because observers fail to recognize the expected characteristics. This might be due to distance, dust, or other factors. If this is the case, then these spirals would not be classified as spirals and would be excluded from the sample set. In this case, the dip in the 80 to 90 degree bin may be due to a selection effect since many appropriate galaxies are inappropriately excluded from the sample set. However, even if these misidentified galaxies were to be included, present methods would be inadequate to determine their inclination relative to the earth if their disk could not be adequately resolved.
That edge-on galaxies should be under represented in the astronomical catalogs because they are more difficult to detect should not be too surprising. In fact, nearby face-on spiral galaxies are still being discovered. Dwingeloo 1 was discovered in 1994. Prior to that time, it remained undetected due to dust, gas and bright stars in our own Milky Way galaxy which obscures around 20% of the sky. What other galaxies does our own Galaxy hide? Only time and further observations at appropriate frequencies will tell.