A FEW NOTES ON THE DARTMOUTH CHIN STUDY

[Chris Bristow]

I am not weighing in on the validity of the chin controversy but after looking closely at  the Dartmouth study I found some problems that may be of interest to anyone investigating their study.
 
The first problem is not obvious because they have placed the mug shot and backyard images on top of each other. They appear to match in size if you compare the width from ear to ear. But because the backyard image is looking very slightly to his left the ear measurement becomes worthless. If you place them side by side you find the eye to mouth measurement is at least 11% larger in the mug shot. The backyard image is looking downward about 6 degrees more than the mug shot, but this does not cause the backyard eye to mouth measurement to shrink and does not affect the comparison.
           (I learned something very non intuitive about this perspective change of tilting the head down. The first 40 degrees of tilt down cause the eye to lip measurement to actually increase a tiny bit instead of shrinking as I assumed. That was a surprise! It turns out as you tilt down you start to see that the eye is farther back on the head than the lips. When the person is facing straight forward you can't measure the fact that the eye sits farther back than the lips. But as you tilt your head down that difference starts to become visible and measurable. I guess portrait artists already know this but it was a new and interesting  aspect of perspective for me.)

The second problem is that the backyard image is looking down about 8 degrees more than the mug shot, but when they created the model they more than doubled the amount of tilt. When you place the model next to the image you find the height of the lighted area on Oswald's chin is smaller in the model. However before you can measure this you have to make the correction explained in the next paragraph 'Problem 3".

Problem 3 is that both computer models are oversized by 6%. Before you make any measurements  you will have to shrink the computer models by 6%. The backyard image is tricky when making this comparison because since it is tilted down more than the photo the dimensions will gradually shrink as you go from top of the face to the bottom.

One last small issue is the model of the mug shot is straight up and down while the mug shot itself leans to the right by 4 degrees. The backyard image is rotated maybe 1 degree more than the mug shot. So for the best comparisons you should rotate all the images to a vertical alignment. 

Both the larger mug shot and the extra rotation downward of the backyard model create a bias in favor of the results they obtained. They both cause the light on the chin to shrink and so the light does not wrap around the side of the jaw as much when the too small model is compared with the oversized mug shot. I tried to resize things and photo shop the corrected chin size onto the mug shot. But because the shadowing on the two images is different it did not blend well. To be fair I did notice the increased size was not very significant but for anyone studying the images I think the information above should be helpful.