Just an FYI Snow, Check Six is a member of this forum. you can reach out to him if you need too?

You are not allowed to view links. Register or Login

Just an FYI Snow, Check Six is a member of this forum. you can reach out to him if you need too?

That's good to know.

But his site serves ads. So it seems he benefits by keeping his list private? Be nice if he wanted to discuss things here. But I suspect it's not his goal to publish his list. Dunno.

As another example, skyjack wasn't motivated to share his list either, because of the time he spent on it. Apparently.

I believe I found my first case where the checksix site incorrectly says a serial is not a match.
But if you look at the image from the fbi file, my determination is clearly the correct one.
So I think all this work is going to pay off. Don't think anyone has a accurate text list.

fbi file 55
This is my correct determination:
H34562085A 1963A page 164 row 41 group C

attached the snapshot.
B is not legal in the 3rd col, so it's definitely a 8. Not a 6. See other 6's there for comparison

Breaking it down per FBI printout:
H34 562 085A 63A

if you plug
H34562085A 1963A

into the checksix site, it says not a match.

Someone can verify this, but I've tried multiple times and think I'm correct. And no, that's a zero, not an O. The editor I use has a slashed-zero font so it's visually more obvious for me.  O/0 fixes are one of the common ocr things I manually override.

I don't have the Tosaw book immediately available, so I can't say what Tosaw listed.
If checksix started with Tosaw (don't know) maybe there's an error in Tosaw also.

you can do the checksix check at You are not allowed to view links. Register or Login
be sure to copy/paste the serial correctly and pick 1963A on the Bill Series dropdown then
click "Check My Bill"

I checked all the Bill Series in the dropddown, and still it failed, so the error is not in the series
He must have some digit/char wrong. I can't tell which though (he doesn't publish his list)

EDIT: I tried a bunch of alternate single char changes but couldn't find any that he said was a match. So it seems non-obvious what the issue is.

Snowmman: See if this is good enough quality, I think there is an upload limit on here.  The writing in the book is my own when I was doing counts a few years ago.

Snowmman: See if this is good enough quality, I think there is an upload limit on here.  The writing in the book is my own when I was doing counts a few years ago.

Excellent. Thank you, fcastle866.
Interesting, the type font in tosaw's book is exactly the same as the Hoover memo. So the Tosaw images are exactly the same as what I showed above.

So I suspect Tosaw essentially photocopied and published it as images.

i.e. Tosaw never did image to text translation...so there's no way for Tosaw to be "wrong" ..but it's not text.
The Tosaw images may be a little better dpi than what I'm working with, but by starting with the fbi memos, there's no question of source...i.e. there's no earlier source than what I'm starting with, I believe.

I'm amazed that apparently the fbi and ha.com never did either. They must have relied on ha.com employees using their eyeballs to interpret the hoover ransom memo images, zooming in when necessary.

Kind of ridiculous at this point. We should all have an open source we all agree on. That's what I'm working on.
I will have it in 3 forms.
1) starting from the 34 pages of fbi images (cropped)

I've got my methodology honed now, so I can see the result will happen with good accuracy.
just will take a week or two more.
Then:
2) the serials matching the images and that format
3) the serials in a single column format with the page/row/group specified.

I will probably have a readme that highlights known typos on the fbi list, and discrepancies with other known semi-public lists like checksix.

automating checks against checksix's site proved to be like getting into fort knox. He's got more blacklist protection then I've seen when automating web access for information from fortune 500 sites.

but: as snowden showed, if there's a way to get something, it will be gotten. You either burn the information totally or you might as well assume it's public information. (nowadays).

You are not allowed to view links. Register or Login

You are not allowed to view links. Register or Login

I believe I found my first case where the checksix site incorrectly says a serial is not a match.
But if you look at the image from the fbi file, my determination is clearly the correct one.
So I think all this work is going to pay off. Don't think anyone has a accurate text list.

fbi file 55
This is my correct determination:
H34562085A 1963A page 164 row 41 group C

attached the snapshot.
B is not legal in the 3rd col, so it's definitely a 8. Not a 6. See other 6's there for comparison

Breaking it down per FBI printout:
H34 562 085A 63A

if you plug
H34562085A 1963A

into the checksix site, it says not a match.

Someone can verify this, but I've tried multiple times and think I'm correct. And no, that's a zero, not an O. The editor I use has a slashed-zero font so it's visually more obvious for me.  O/0 fixes are one of the common ocr things I manually override.

I don't have the Tosaw book immediately available, so I can't say what Tosaw listed.
If checksix started with Tosaw (don't know) maybe there's an error in Tosaw also.

you can do the checksix check at You are not allowed to view links. Register or Login
be sure to copy/paste the serial correctly and pick 1963A on the Bill Series dropdown then
click "Check My Bill"

I checked all the Bill Series in the dropddown, and still it failed, so the error is not in the series
He must have some digit/char wrong. I can't tell which though (he doesn't publish his list)

EDIT: I tried a bunch of alternate single char changes but couldn't find any that he said was a match. So it seems non-obvious what the issue is.

Snowmman: See if this is good enough quality, I think there is an upload limit on here.  The writing in the book is my own when I was doing counts a few years ago.

PDF instead, may be higher quality.

Apologies for three posts in a row.  There are a good amount of star notes listed in Tosaw's book. I did notice that many of the bills had star notes in sequence, but the bills themselves were not in sequence.  Meaning that note 12345678* was followed by note 89783457*.  There are hundred of star notes in the book.  It gets tedious counting those.  I assumed there were only a few star notes.  All of this is in hindsight, but the FBI could have narrowed their search for these bills by just focusing on star notes of specific years.  So instead of asking a teller to look at a list of 10,000 notes, maybe they could have said look for star notes of 1963A from this Federal Reserve bank.  It would have been easier to search for that needle in the haystack.  More needles, smaller haystack.  Or, Cooper could have seen this himself and decided to spend the bills with the stars and erase the stars.

If someone has the time, they can copy the pages and scan those and then put those into Excel. I'm guessing that's what the folks did that have the info.  I guess its possible that Check Six did it by hand.  That's a lot of typing.

If someone has the time, they can copy the pages and scan those and then put those into Excel. I'm guessing that's what the folks did that have the info.  I guess its possible that Check Six did it by hand.  That's a lot of typing.

I think you don't see what I'm saying. Copying the pages doesn't give you the text, like as if you typed it in.
that's what makes it tedious. Even with the higher quality Tosaw, ocr will give you errors. It's all about verification. Context per page helps with a lot of verification. I have some other things I do that rely on context. But visual is the final verification.

I don't think many people have a list. I don't think ha.com has a list, except for what they wrote down that they verified.

in terms of describing the scale of the problem:
Each serial is 3 + 3 + 4 + (2 or 3) chars.  (the 4 might be legal to be 3, but I've not seen it in this list)

so that's 12 or 13 characters per serial that have to be exactly right.
Multiply that by 9998 and you have between 119976 and 129974 characters to deduce in total.
And you want 0 errors.

That is better than a 1 in 119976 error rate. i.e. 0.0008335% error rate.

EDIT: the context that I'm talking about that helps is:
We know each page is sorted
We know there aren't duplicate serials
We know there is a restricted set of legal serials. All chars are not random. There are positional limits on choices within each of the 12 to 13 chars). And the series year choices are fixed. Additional the leading letter range is just A-L (fed reserve choices) and the trailing letter in the serial is just A or B or C or D or *

(yes I have seen some C and D ..still verifying though. mostly A or B or *

valid series year are "34", "50", "50A", "50B", "50C", "63", "63A", "69"

interestingly the 1934 twenty is dramatically different looking..has different words on it.

if the chars in the serials were fully random it would be a harder task. Also if the fbi list wasn't sorted it would be a harder task. I suspect the sort also helped fbi's verification process, whatever it was.

You are not allowed to view links. Register or Login

If someone has the time, they can copy the pages and scan those and then put those into Excel. I'm guessing that's what the folks did that have the info.  I guess its possible that Check Six did it by hand.  That's a lot of typing.

I think you don't see what I'm saying. Copying the pages doesn't give you the text, like as if you typed it in.
that's what makes it tedious. Even with the higher quality Tosaw, ocr will give you errors. It's all about verification. Context per page helps with a lot of verification. I have some other things I do that rely on context. But visual is the final verification.

I don't think many people have a list. I don't think ha.com has a list, except for what they wrote down that they verified.

in terms of describing the scale of the problem:
Each serial is 3 + 3 + 4 + (2 or 3) chars.  (the 4 might be legal to be 3, but I've not seen it in this list)

so that's 12 or 13 characters per serial that have to be exactly right.
Multiply that by 9998 and you have between 119976 and 129974 characters to deduce in total.
And you want 0 errors.

That is better than a 1 in 119976 error rate. i.e. 0.0008335% error rate.

EDIT: the context that I'm talking about that helps is:
We know each page is sorted
We know there aren't duplicate serials
We know there is a restricted set of legal serials. All chars are not random. There are positional limits on choices within each of the 12 to 13 chars). And the series year choices are fixed. Additional the leading letter range is just A-L (fed reserve choices) and the trailing letter in the serial is just A or B or C or D or *

(yes I have seen some C and D ..still verifying though. mostly A or B or *

valid series year are "34", "50", "50A", "50B", "50C", "63", "63A", "69"

interestingly the 1934 twenty is dramatically different looking..has different words on it.

if the chars in the serials were fully random it would be a harder task. Also if the fbi list wasn't sorted it would be a harder task. I suspect the sort also helped fbi's verification process, whatever it was.

I think we are talking about different things within the same concept.  We are using the list of bills for different projects.
 In my research I would have found it useful to have an Excel spreadsheet with all the serial numbers.  Useful may not be the term, more like I would have liked to have had it, like I like to have books on the case or memorabilia, etc.  So, I took Tosaw's book, copied a page, then scanned it to PDF, then exported it to Excel.  It worked.  However, I did not want to make the effort to go through all 60 pages.  Granted, I did not go line by line to verify the error rate, so that could have been an issue for some.  However, for me, I was looking to compare the number of 1963A to 1969s, and the number per Federal Reserve bank.  I ended up doing this all by hand.  I didn't need to be 100% accurate.  I got it done to fit what I wanted.  Every now and then someone asks to check a serial number or do some money research (as a hobby).  Having a list on Excel would be useful for them.  It would have helped me.  But, like I said, I figured it out for what I needed.