Each of these is the result of taking text from two portions and overlaying them. The process for each is, using Paint Shop Pro 8:
1. Invert each portion (negative image)
2. Reduce the luminance by 50% so that it never gets brighter than 50% gray.
3. Colorize one with 100% red (hue=0, saturation=255)
4. Colorize the other with 100% blue (hue=160, saturation=255)
5. Composite the image with the blue image on the bottom at 100% opacity and the red image on top at 50% opacity.
IMAGE #1:
- RED = line 1a
- BLUE = line 8
- This looks okay to me...I would expect the blue text to be narrower (the curvature results in a narrower image in the projection plane)
IMAGE #2:
- RED = line 1c
- BLUE = line 8
- The close horizontal alignment tells me that the projection geometry is (most likely) nearly identical for those two regions on the page. Measuring the widths of lines 6a and 13, respectively shows a difference of approximately 8 pixels, so I should expect a trapezoidal effect on width varying with vertical position on the page, but the horizontal metrics on the right half should be fairly consistent.
IMAGE #3
- RED = line 11
- BLUE = line 9
- I'm not concerned with the difference in the baseline positioning between uppercase "A" and the rest of the word for the two images. Using the shift key on a manual typewriter physically moves the entire carriage, so there shouldn't be any expectation of consistency. The difference in horizontal width, however, suggests that the left half of the page differs considerably from the right half in the projection plane. This isn't really a red flag, and the fact that the delta for each successive letter grows consistently is what I would expect.
IMAGE #4
- RED = line 6a
- BLUE = line 7c
- I chose the 6a "Honolulu" instead of the 7a "Honolulu" specifically because it has a similar horizontal position to line 11 "Africa". This means that IMAGE #3 and IMAGE #4 should have very similar horizontal characteristics. However, this image is very strange. The overall width of "Honolulu" is the same, but the letter spacing is inconsistent. I can't claim forgery, but it is definitely anomalous.
I also looked at a few other things, but nothing jumped out at me as red flags:
- Looked at the rotational alignment of capital "H" since it seems like it is rotated clockwise relative to other letters. The rotation appears consistent throughout the document.
- Looked at the top serif of lowercase "l" since it extends beyond the bottom serif. Where the top serif is completely present, the overshoot is consistent and it doesn't appear to be kerned with the preceding letter.
- Looked at the height of lowercase "l" for consistency.
- Looked for consistency in chromatic "ghosting". Red/blue ghosting is not uncommon and is a normal result of diffraction effects in the scanning optics. Looked specifically for consistency in chroma ghosting for various items based on position in the document.
