Analysis Of 18,000 Random Bills In Circulation

I've always wondered what I would find if I did a large-scale analysis of dollar bills in circulation. So...Here's what I found! Questions and comments welcomed!

This is the scope of my study:


Analysis of Series:


--Comparing SJ23 and SJ24 shows the reduction of 2009 and 2013 by about 30%, and reduction in 2017 of about 24% as the new 2017As and 2021s made their way into circulation. 2003 and earlier were 1.2% of the total.
--Just looking at the 2023 samples in SJ23 and NJ, 2021 had not made their way into circulation yet. There's a remarkably even distribution of 2017A, 2017 and 2013. 1.4% of the bills were 2003 or earlier.
--A few months later, 2021 represented 3.5% in BOS and 8.5% in ATL. BOS is not on my chart because I only had limited time there, enough only to count 2021, not other series.
--I noted a significant number of crisp 2017A in both SJ23 and ATL, not perfectly sequential but looked like they had just been put into circulation although they were printed October 2022.

Production:

--Washington DC production is about one fourth of total BEP production going back to 2009.
--Just under 50% of the bills I found in NJ and BOS were DC production, so about twice the national average.
Only 12% of what I found in SJ was printed in DC, or about half the national average. That 12%, I'd assume, migrated to SJ rather than being distributed there from the BEP.

Federal Reserve Letters:

--The green boxes are the 'home' FRD for my sample locations
--There's a strong correlation between FRD letter and Location for both ATL, SJ23 And SJ24, all about about 2x the national average for their own letter.
--For BOS, I was wondering if I'd see like 50% FRD-A but only found 9.2%, still 1.7x the national average but I wonder why not more. They're not in neighboring New York, nor are there many FRD-B found in Boston.

Late block letters and star notes:

--On average, I found 33.5 late-block bills per 1000-bill brick. Other than S, there was a steady decline in the count of each letter as I got deeper into the alphabet. I found a single Y-block, that was a big "I wonder" for me.
--I found significantly more late blocks in SJ. Late blocks are strongly correlated to FRD-L, so the fact I found 2x the national average FRD-L in SJ23 and SJ24 explains why I got so many in late blocks.
--Note the decline in almost every letter between SJ23 and SJ24 The majority of late blocks I found were SR 2009, and as 2009s are taken out of circulation so too go the late blocks. The 30% percent decline in late blocks in SJ matches almost exactly the overall drop in 2009 bills.
--Star notes in my sample were pretty evenly distributed across locations but way less frequent than the production rate of 1.28%, or 1 in 78. I only found one in 140. That's a big difference for a large sample.

Interesting finds:

--1,2: the lowest and highest serial numbers I found.
--3,4: a repeater and four pairs (both ~1/10,000). I found zero radars.
--5: FRD-I star note out of a production run only 250K
--6: The one and only Y-block I found.
--7-10: Overinking on the 3d digit of the 2013s has been discussed on here before, that's #7. But #8 is 2017A, totally different series! And note the suggestion of overinking on #9 and #10 from 2021. I noticed far more 2013 overinking in my 2023 sample, maybe 1 or 2 per strap, but hardly any in 2024.
--11 I was wrong when I said bill #2 was the highest, but I had assumed 96M was the highest serial number. I found this and it's legit for this series, as pointed out by SteveinTampa, thank you, that's actually cool.
-12. this >96M bill was cut from a sheet (badly, may I add) and then circulated.
--I saw no errors and nothing earlier than 1963, no silver certificates.
--I was desperately hoping to find 5 consecutive 9's (odds: 1 in about 38,000) because I have five consecutive of everything else. Gotta keep trying I guess!

I'll be adding to this with an analysis of production timing and distribution, which is a little more complex...until then questions and comments welcomed!