The Andrew Turnbull Network

Quick navigation

[Our only earth] Atheism
[Road sign] Traffic Signs
[LOL-505] Licence Plates
[shopping cart] Supermartifacts
[Third Eye Blind] Lost A Whole Year
[Mozilla] Mozilla Network
[map] Mercer County
[Walk the Moon] The Tightrope

Get SeaMonkey!
Thank you for not using Chrome.

Welcome to the personal website of Andrew Turnbull. This outpost features tons of stupefying and trivial things pertaining to various and diverse interests of mine. Chances are, if there's something I know about or like that doesn't much other representation on the 'net...there's a bit of it here.

April-June 2022 Archive

14 June 2022

When licence plates run out of numbers...

For the first time since the start of Endless March and the pandemic, I've had an uptick of interest in licence plates. No, I don't have any collection additions to share. No, I don't think I'm going to make it to the 2022 ALPCA Convention (unless I change my mind, register late, and dare a trip across the's still a month away). But I am back in the mode of observing and researching the things again...which is probably what I do best.

When boiled down to their fundamental essence, licence plates are essentially sequences of letters and numbers that progress through the alphabet in tandem with the progression of time. The two most essential online resources that track these are Eric Tanner's All About License Plates, and Kit Sage's High Road. Several collectors such as Jim Moini in New Jersey also track highs and cutoffs in their particular areas of influence. I did some of that myself, until the relationship between Wisconsin and I became torched beyond repair.

A two-letter, four-number format has a capacity of 6.76 million combinations. A three-letter, three-number format has capacity for over 17 million. That's a lot. But not even the largest-capacity format has enough capacity to last forever. Old formats are constantly running dry, and new formats are constantly appearing to replace them. Here are eleven places where this will happen in the months and years ahead:


[licence plate] A new format in California doesn't happen very often...and when it does happen, it's a Really Big Deal. The current 1AAA000 format dates back all the way to April 1980. The leading digit advances after the letters, and California advances to a new digit every four to six years.

Trouble is, the format is running out of numbers. The state turned over from 8ZZZ999 to 9AAA000 earlier this year, and is presently in the "9BV" series. It's no cause for alarm, though. Following past precedent, California will likely reverse the format to 000AAA1 when 9ZZZ999 is reached. This will give the state enough capacity to last another four decades, more or less.


[licence plate] Idaho's plate numbering is mighty idiosyncratic. Rather than use a single statewide series, the state assigns each county an alphanumeric code and the will to go through licence plate numbers at its own rate. Even more idiosyncratic is the fact that the codes are alphabetical mnemonics: The first county that starts with "B" has code 1B, the second is code 2B, and so forth. For the seven counties that don't share their leading character with any other, lone letters are used. Seems inefficient, but good luck changing an entrenched system in a predominantly-rural state where even the slightest hint of progressive change gets white people up in arms.

That's not the most inexplicable thing about Idaho's numbering, though. In late 2020, the state changed the format of its plates to implement a constant "U" suffix after the number. (Why U? Who knows.) Since the county code already takes up two of the available seven characters, that means that plate serials are now effectively limited to four digits.

This means utter headaches for counties like Ada, Canyon, and Kootenai, each of which have far more than 9999 vehicles in their areas of control. It's also a headache in Butte County (code 10B), where serials are now limited to three digits thanks to its outsize code. The result? So far, a mad flurry of new alphanumeric formats. Ada County has already torn through 1A A001U through 1A Z999U, 1A AA01U through 1A ZZ99U, and part of 1A AAA1U through 1A ZZZ9U in the last year and a half. What could be next? They could continue to pile on letters, and end up with 1A AAAAU through 1A ZZZZU...which could yield some interesting five-letter combinations (1A FUCKU?) if the state isn't vigilant screening them out. But where Idaho is going with this is anyone's guess.


[licence plate] Ah, Maine. The current alphanumeric progression started at 1 AB in 1991. 32 years and one reissue later, they're in the "ZF" series...and plate number 9999 ZY is only a year or so away.

In the (distant) past, whenever Maine ran out of numbers, they would pile letters onto the format. 999-999 was followed by 1 A in 1979, and 99999 Z was followed by 1 AA in 1990. By that precedent, the next format to follow 9999 ZY should be 1 AAA through 999 ZZZ. There's a catch, though: Since 2003, this format has already been in use on various special-issue plates. Regular passenger plates could easily pick up dedicated blocks in this format, though...and there's precedent for this, as other New England states have moved towards using a single alphanumeric format for multiple plate types in recent years.


[licence plate] Nebraska is a state with a split personality. Since 2002 the Lincoln and Omaha metro areas have issued plates in a straight AAA 000 format, with each baseplate picking up where the previous one left off in the sequence. The rest of the state (making up a minority of its population, in spite of hoarding most of the land) soldiers on with numeric county codes and formats that have been unchanged since the 1960s.

The Lincoln/Omaha series is skirting dangerously close to the end of the alphabet. Serials hit YAA this year...and with a reissue scheduled for 2023, they're going to burn through a lot of numbers soon. If the official mock-up of the state's next baseplate is any indication, though, Nebraska plans to simply wrap around to the beginning of the alphabet. The state has capacity to spare, and the entire A through M series have never been used at all.


[licence plate] Nevada's current baseplate commenced issue in late 2016, with numbers starting at 001-A00. Five years on, the state has reached the "W" series. Z-series plates are reserved for dealer-processed registrations, so the presumable final number of the format will be 999-Y99...and only a few months away.

Nevada has a penchant for occasionally backfilling unused portions of formats, leaving nothing wasted. After plate number 999-YZZ was reached in 2013, the state ran through portions of the L and A series that had been skipped over in 2001 and 1984. The next format to follow was 00A-001...but this series only got through mid-H before being suspended in 2016 due to the advent of a redesigned baseplate. With this in mind, my guess is that Nevada will retrace its tracks and backfill 50H-000 through 99Y-999.

New Jersey trailer

[licence plate] New Jersey is an interesting state for licence plate spotters. The last year it engaged in a general plate reissue was 1960...meaning that 62-year-old plates could still conceivably be on the roads (though since plates transfer with the owner, not the vehicle, anyone driving on them is going to be an octogenarian at the very least.) This also presents an issue for choosing alphanumeric formats, since the state must prevent duplication with every past format of the last 62 years! Nevertheless the state has coped with the situation well, sticking to six digits in formats like AAA-10A that aren't too convoluted or unreadable.

New Jersey's current passenger series started at A10-AAA in 2010, and is currently halfway through its sequence. Trailer plates have traditionally used many of the same formats as passenger, but with a catch: The key serial letter is always "T."

The current trailer series began at TAA-10A in late 2004. Seventeen years later, the sequence has reached the "TZF" series...and a new format is around the corner. What could be next?

My guess is that TZZ-99Z will be succeeded by A10-TAA, following the current passenger format. (Note that the leading serial letter advances before the fourth through sixth characters.) The passenger series should skip from Z99-SZZ to A10-UAA when it reaches this point in the alphabet, also in the near future.

South Carolina

[licence plate] South Carolina's current passenger series began at AAA 101 in 2008. Fourteen years, one full reissue, and one rolling replate later, the series has inched up to "VQS." The end of the series is probably two or three years away.

What will happen after that? Logically, the state would reverse the format to 101 AAA and run through it for the next 15 to 20 years. This sequence was last used on South Carolina plates between 1998 and 2008; none of which are still in use.

West Virginia

[licence plate] My often-poignant, often-difficult home state is also difficult on the licence plate format front. The first character of all passenger plates is a month code keyed to expiration...meaning that there are twelve different series that progress in parallel. Using July as an example, the state ran through 7A 1001 to 7Z 9999, then piled on a letter and used 7AA 101 to 7ZZ 999. So far, so good.

When some of the two-letter series approached the end of the alphabet in the mid-2010s, I expected the state to pile on yet another letter and issue 7AAA 11 through 7ZZZ 99. Instead, the state caught me by surprise and began using a modified single-letter format: 71A 101 through 79Z 999, with the second character being a numeral that advances after the letter. The first character remains linked to the month, as before.

As of 2022 West Virginia has reached the "78" and "O8" series, with more months sure to follow. Another format is only a year or two away. Then what?

We're in uncharted waters here, and it's impossible to say for sure what will happen until it happens. My guess is that the state will continue floating the serial letter around to different positions, and 79Z 999 will be followed by 711 A11. But the state might choose a different format, or switch to a single series and scrap the month coding altogether.

British Columbia

[licence plate] North of the border, Canada's west coast province has been resistant to reissues since 1986. All general-issue plates issued in the last 37 years remain in use...with the province coping with serial format exhaustion by periodically rearranging the numbers and letters. The current passenger series started at AA0 00A in 2014, and is currently up to "SB." Although jumbled, this format is actually a marvel of logical sequencing that laid to rest the confusing split-alphabet format that BC had inexplicably stuck to for the 44 years up till then. The next format will probably arrive mid-decade...and by past precedent, I'd expect it to be a reversal (i.e., A00 0AA) of the one they're currently using.

BC's passenger plates aren't the only ones with a format inching towards exhaustion, however. Motorcycle plates are destined to run out of numbers first...and here, the suspense runs high.

[licence plate] All British Columbia motorcycle plates back to the late 1970s have used a one-letter, five-number format. At first, this letter ran through in reverse sequence: The 1974 base used N, the 1979 base used M, L, K, and J, and the 1986 base picked up at J, followed by H. But a year or two into the current series, all sense of logical sequencing was thrown out the window. H was followed by N...then the province went through (or reused) C, M, E, K, L, S, U, W, Y, and Z in that order. Z99999 was followed by a backfill of the low J series, bringing the series right back to where it started 37 years ago. Every prefix in the rest of the alphabet is already used by another BC plate motorcycle plates are plumb out of wiggle room, unless they switch to a new format.

What will the next BC motorcycle format be? Following the precedent of other types, I expect the A00000 format will be followed by 0A0000...using random, unpredictable, and out-of-order serial letters as before. It wouldn't be like BC to not be exasperating.

Ontario trailer

[licence plate] Ontario's current passenger series started at AAAB-001 in 1997. The province's format allows for an absurdly high number of combinations, and isn't likely to run out of numbers for the better portion of 200 years.

Trailer plates aren't quite as well-set for the future, however. The current format started at A10-01A in 2002, and is presently up to the W-Z series. Plate number Z99-99Z is only a couple years away.

In the past, Ontario coped with trailer plate format exhaustion both by reversing the format (999-99M to A10-001) and by piling on letters (Z99-999 to A10-01A). The current format is symmetrical, so it obviously can't be reversed...but they could add a letter yet again, and follow Z99-99Z with AA-001A or A001-AA.

Another possibility is that the province will follow the lead of truck plates, and squeeze another digit into the existing configuration: A10001A. This would be far from ideal: Trailer plates have a vertical legend, and there's simply not enough room on them for a seven-digit serial with a space. That didn't stop Ontario PRP plates from overflowing from 999-9PZ to PA10001, though...


[licence plate] Lastly, we have la belle province du Québec. Quebec is even more resistant to licence plate change than BC is: Its last reissue happened in 1979, and its baseplate (which is still fully embossed!) hasn't been redesigned or tweaked since 1983.

Even the most fervent resistance can't stop formats from running out of numbers, however. Quebec's current alphanumeric configuration is its fourth since 1979, and started at B01 AAA in 2010. The province has torn through letter series at a rate of one or two per year, and is currently up to "ZCQ." As in New Jersey, the leading letter advances before the last three.

Logically, Quebec ought to follow past precedent and follow Z99 ZZZ with a reversed format starting at AAA 01B. The official word, however, is that they'll instead rearrange the letters and start the next format with 01B AAA. (Or 01E AAA: By some accounts, all the letters from A through D are skipped in the first position now.) Why? Why not.

2 June 2022

25 May 2022

A Musical Mystery

[Andrew at WVU, 2005]

Way back in the summer of 2005, a younger version of myself was living in Morgantown, West Virginia, trying to survive the nadir of the Bush years. There was little joy in my life, but the second-to-last weekend of August did give me something to look forward to: Fall Fest, a free concert for the WVU student body.

I was there. Most of the crowd was psyched about the headliner of the evening: A band called O.A.R. I didn't care about O.A.R. I was more interested in the opening bands...since this was where the offbeat talent on the bill, the stuff I was actually attracted to, was!

The show began. The very first opener on the bill was a band called Filomath. They performed several dozen minutes of melodic guitar-driven rock music, à la Oleander or Lit. Between songs, the singer excitedly told the audience "We've been signed to RCA!" and talked about the flurry of activity that was sure to follow. After their set, the members retreated to the merchandise table to greet fans and sign autographs. I thanked Filomath for the show and bought their CD on the spot, sensing that they were an up-and-coming band on the eve of a big breakout.

I never heard the group again. Their website went down around 2007, though their MySpace page hilariously still exists.

Whatever happened to Filomath? What became of their contract with RCA? And for that matter: What WAS Filomath in the first place?

[Filomath CD]

Well, it wasn't a dream. The CD booklet lists the members of the band as Leroy (Pelicci), Lee Nadel, Markus Dorfmann, and Mike Reid. Were these the same as the musicians who jotted the chicken scratchings shown above? Who knows.

There's almost nothing of record about Filomath on the Internet: No news articles, no biographies, no detailed discographies. Discogs, which has a listing for virtually every microvariation of every piece of recorded music on earth, has a skeletal page for the band that doesn't even list their one album. Concert Archives says a little more, but they list only one concert that Filomath is known to have performed: The very same Fall Fest, in August 2005.

Digging deeper, the plot thickens. Discogs doesn't list any album by "Filomath," but they do list an album by a group called "Carbondale" that contains the exact same songs in the exact same order. The Carbondale and Filomath albums also have the same musician and production credits, and even have the same barcode. The Filomath CD bears a 2004 copyright, but the Carbondale CD came out in November 2003. Clearly, Carbondale became Filomath. Was there a legal issue that forced this band to scrap its original name? ...Who knows.

Did the members of Filomath accomplish any other musical pursuits? Yes. Discogs credits half the band as contributing to obscure Deridian and Synapse records in the 1990s, and this nice throwback-style website sheds some light on Pelicci's past as a part of the Dick Fawcett Band (while also iterating that he "went on to record for RCA Records"). Trouble is, all these musical projects happened before Filomath, not after. Have the members recorded or performed at all since that fateful day in August 2005? ...Who knows.

And with that, the trail goes cold. Maybe the answers are lurking in the dark web of Facebookstagram, Twitter, or some other hellsite I won't touch except with a 3.05-metre pole? ...Who knows.

[Filomath CD]

Maybe someday, I'll have the answers. But for now, the fate of Filomath remains a musical mystery.

18 May 2022

Before I die, I want to hear a Catholic/Evangelicalist/Fundamentalist Christian (my far-right antisemitic Lutheran cousins in Wisconsin, say, or any of the Jesus-soaked Bush-shilling warmongers I went to high school with in West Virginia) admit three things in my presence:

As usual, I'm not holding my breath.

11 May 2022


[RSS icon] If you eyeball the sidebar on the left, or scan the location bar in SeaMonkey, Pale Moon, or any other browser worth a damn (which patently excludes Firefox and Chrome these days), you'll notice an icon that wasn't there before: An icon to an RSS feed.

You can view an RSS feed in your e-mail client, or as a Live Bookmark (again, in SeaMonkey, Pale Moon, or any other browser worth a damn.) So you can keep tabs on the pointless drivel I write, without even having to visit the website:

[Live Bookmarks]

Like everything else I create, this file is a hand-written static document. I'd like to write a PHP file that automates RSS creation for me...but programming isn't my strong suit, and I still haven't figured out how to do it.

And good luck learning how. On the Internet, innocuous questions like "How do I create RSS using PHP" attract responses like this or this: "You shouldn't write your own code! You should instead blindly link to massive, bloated libraries of other peoples' code without understanding it, and I'll passively-aggressively belittle anyone who thinks otherwise." Which sums up everything that's wrong about web development as it exists right now.

7 May 2022

Code Cracking: On the Bumper

No, I didn't make it to the Acton Licence Plate meet on April 30th. Thunder Bay, Ontario is as far away from Acton as Fredericton, New Brunswick is...and getting there and back would have required four days on the road, and five tanks of gas.

Licence plates have lost a lot of lustre lately. Part of this comes from living in a post-Michael Brown, post-George Floyd world where it's common knowledge that police departments are insubordinate, out-of-control institutions, dominated by violent racists determined to extinguish any and all public trust of their profession. And when cops between outbursts aren't systematically scanning plates for personally identifiable information, far-right legislators in US states are using them to convey official endorsement of exclusionary, theocratic positions...and collectors collect them, to flaunt their own bigotry.

The 3M Company still pushes its proprietary "Digital License Plate technology" to governments and prisons (which still make licence plates, for some reason), with Ohio being the latest state to fall. Suffice to say, I'm not a fan of this trend either. But "digital" plates did facilitate one development that's worth exploring: The advent of barcodes on licence plates.

[Barcodes on licence plates]

Barcodes like these are basically useless in the field: No matter how overzealous cops may be, it'd be a stretch to imagine one crouching behind each and every car with a scan gun! So why are they on plates at all? To make life easier for the DMV employees doing data entry at issue time, that's why. All these barcodes are Code 39, a common and well-documented format developed in 1975 and used on everything from asset tags to VIN plates.

I'm unclear what the very first licence plate with a barcode was: Some embossed Mexican plates had them as early as 1998, and the Canadian province of Newfoundland experimented with barcoded validation stickers around the same time. Since then, barcodes have made appearances in at least 17 U.S. states, Nunavut, and Prince Edward Island. Unsurprisingly, all but one of these places issue flat plates.

What data, precisely, do these barcodes contain? I wanted to I scrounged a used barcode scanner, rummaged through my collection and trade box, and found out!


Five states: Barcode contains plate number, and nothing else.

Well, that was anticlimactic! But now you know:

Licence plate: Information encoded in barcode:
2014 Arizona, #BDC4264 BDC4264
2015 Georgia, #PEL9205 PEL9205
2015 Georgia, #PSD9879 PSD9879
Undated Georgia, #PYR4245 PYR4245
2013 Minnesota, #912-KGH 912KGH
2014 Oklahoma, #889KYO 889KYO
Undated Texas, #BPX-8483 BPX8483
Undated Texas, #FKW-1974 FKW1974

Five states: Barcode contains plate number, plus an additional code.

The plot thickens. The Iowa code is deceptively simple, yet perplexing: Could "1" be locally-issued and "2" centrally-mailed, or vice-versa...or something else entirely? The Alabama PC and South Carolina RP1 might be a class code...or it might not. Tennessee's "1000" definitely is a class code, and "6" is the last digit of this baseplate's year of initial that's the one barcode of the bunch I feel fairly confident about. The Nebraska plate's barcode has an arcane string of letters and numbers printed next to it...and the "LD11" designation, like Tennessee's, appears to represent the baseplate's year of initial issue.

Licence plate: Information encoded in barcode: Notes:
2015 Alabama, #47CN390 47CN390 PC
2005 Iowa, #246 LTY, Linn County 246LTY2
2009 Iowa, #973 TLM, Butler County 973TLM1
2015 Nebraska, #TPG 411 TPG 411 LD11 Code "01DWT LD11 0121937" is printed next to barcode.
Undated South Carolina, #LEQ 404 LEQ404 RP1
2015 Tennessee, #P95-67G 10006P9567G "Class Code 1000, Issue Year 2006, 10006P9567G" is printed opposite corner from barcode.

Two states: Barcode contains a unique serial not connected to the plate number at all.

Here, the plot gets thrown out entirely. Most flat Indiana plates are emblazoned with a 12-digit number that inches upward in sequence...but it couldn't be a literal serial ticker of plates made, unless a state with fewer than 6 million vehicles somehow spent ten years issuing ten plates to every one of them! Meanwhile, the Idaho code is a complete mystery that defies explanation:

Licence plate: Information encoded in barcode: Notes:
2014 Idaho, #1A 9B773 PL3266079
2013 Indiana, #855BHC, county #3 000048700237 12-digit serial is printed next to the barcode.
2014 Indiana, #846TTN, county #41 000059482062 Last 5 digits of encoded number (82062) is printed next to the barcode.

I'd love to explore the mysteries of these barcodes in more detail. Unfortunately, without having a pile of surplus licence plates on hand, it'll be difficult to do any more research on this front.

1 May 2022

Code Cracking (addendum)

What have I learned since dipping my foot into the world of proprietary publishers' barcodes a week ago? Quite a bit, actually!

First off, we can diffuse the "mystery" of the first bit in each 5-bit group. There's no information hidden here: It's just that for whatever reason, Dell Publishing chose to prefix the binary values of half the numbers with "0" and the other half with "1:"

Value Encoded as
0 00000
1 10001
2 10010
3 00011
4 10100
5 00101
6 00110
7 10111
8 11000
9 01001

This was consistent: For example, 7 was always encoded as 10111, never 00111, regardless of what position it appeared in.

[Dell ISBN]

Like the rest of the publishing industry, Dell adopted the International Standard Book Number in the early 1970s. Their four-digit catalogue numbers were adapted to the new format through the addition of extra characters: 0462 became 0-440-00462-4, for example.

Initially, the fifth digit was "0" on all Dell ISBNs. About 1977, however, the publisher began using the fifth digit to indicate different imprints and series: "4" indicated Yearling, "9" indicated Laurel Leaf, and "1" and later "2" indicated plain-Jane Dell.

How did this affect Dell's barcodes? To accommodate the extra digit, the central "division" character became linked to it. The Beatles Lyrics Illustrated (ISBN 0-440-90615-6, list price $1.75) carried a barcode number of 061590175.

[1970s Dell paperback barcode]

Lastly, Dell Publishing used its barcoding system for a far longer period than I expected: Twenty years, to be precise! The above paperback is a 1989 printing of C.T. Westcott's Broadsides & Brass (no, I didn't read it). For those keeping score, the proprietary Dell barcode number for this book is 024020350. The code appears on the inside front cover alongside the more modern EAN "Bookland" barcode, and an OCR-A ISBN and price code string. All three of these identifiers contain essentially the same information: ISBN 0-440-20240-X, list price $3.50.

23 April 2022

Cracking the Dell (Bar) Code

[1970s Dell paperback barcode]

Last week, I paid a visit to the Thunder Bay Public Library book sale. I picked up a highly-questionable pseudo-diet tract off the "free" paperback shelf, and brought it home with me...for no reason other than the fact that I was intrigued by the strange pattern inside the front cover (above). What could this conglomeration of bars mean?

It was quite evident that this was some type of barcode for machine scanning, not unlike the familiar UPC and EAN symbols that are on books (and virtually every other consumer product) today. It obviously wasn't a UPC or EAN, though. It was an early, proprietary format of barcode exclusive to the publisher: Dell Publishing, the New York-based pulp purveyor who ran this title off their press in May 1973.

Did any documentation of this code exist on the Internet? An online search brought me one lead: This thread on the Chronicles science fiction/fantasy forum. Some readers there had noticed similar barcodes on Dell paperbacks with publication dates as early as July 1969...and were as befuddled by them as I.

Laying out the evidence, I made the following observations:

  • The device consisted of two vertical groupings:  One at left consisting of 45 bars, and one at right with a variable number of bars and spaces.  It was obvious that the right grouping contained encoded information, while the left grouping simply served to align the scanning machine.

  • It was probably a simple code...after all, it had only a single bar width, and had to have been capable of being read by primitive, 1960s-era computers.  It also predated the advent of more sophisticated coding systems, including UPC.

  • Since a decimal digit requires at least 4 bits to be expressed in binary form and 45 is divisible by 5, it was plausible that each group of 5 bars or spaces was equivalent to 1 decimal character.

  • Judging by the alignment, the code was likely read from top to bottom.

  • Following the precedent of modern "Bookland" EANs, the information contained in the barcode likely consisted of the book's catalogue number and list price.

Could I crack this code myself? Let's try! Armed with four sample codes (one from the book in my hand, and three from the Chronicles thread), I converted the bars to binary strings under the presumption that a bar equalled 1 and a space equalled 0:

Dell paperback barcode

Exploring the "each grouping of 5 bars or spaces is equivalent to 1 decimal character" theory, I broke the binary strings into pieces and converted them to their decimal values:

Binary Decimal
10001 11000 00000 00101 10001 00000 10001 00101 00000 17 24  0  5 17  0 17  5  0
00000 00011 00000 10111 10001 00000 00000 10111 00101  0  3  0 23 17  0  0 23  5
00000 10100 00110 10010 10001 00000 00000 00101 00000  0 20  6 18 17  0  0  5  0
10111 00110 00110 00011 10001 00000 00000 00101 00000 23  6  6  3 17  0  0  5  0

Immediately, two things jumped out to me: First, all four books had the values "17 0" at the midpoint of their strings, indicating a division of some kind. And although the two-digit numeric values were nothing more than noise, the single-digit values lined up with the books' catalogue numbers and list prices!

Value Cat. no. List price
17 24  0  5 17  0 17  5  0 1805 1.50
 0  3  0 23 17  0  0 23  5 0307 unknown
 0 20  6 18 17  0  0  5  0 0462 unknown
23  6  6  3 17  0  0  5  0 7663 unknown

Going back, I discarded the first bit of every 5-bit group and converted the second through fourth bits into decimal.  Suddenly...everything clicked!

Binary Decimal Cat. no. List price
0001 1000 0000 0101 0001 0000 0001 0101 0000 1 8 0 5 1 0 1 5 0 1805 01.50
0000 0011 0000 0111 0001 0000 0000 0111 0101 0 3 0 7 1 0 0 7 5 0307 00.75
0000 0100 0110 0010 0001 0000 0000 0101 0000 0 4 6 2 1 0 0 5 0 0462 00.50
0111 0110 0110 0011 0001 0000 0000 0101 0000 7 6 6 3 1 0 0 5 0 7663 00.50

The code contains nine digits of encoded information:  The four-digit Dell catalogue number, a "1," and a four-digit list price in US cents.  Pretty much what I expected it to be.

But why was it printed on the book in the first place?  In 1973, retailers weren't set up to handle barcode scanning.  Since this was a proprietary barcoding system not shared with any other publisher, it would have been fanciful to expect these codes to have ever been used for this purpose.  But a 1975 news article (also linked to in the Chronicles thread) explains the reason why:

"When a mass-market paperback is not sold, the dealer and distributor don't return the whole book to the publisher for credit, but only the front cover. At Dell's distribution center in Pine Brook, N.J., returned covers are run, at almost the speed of light, through a scanning machine to compile such information as number of copies sold of each title, royalties due authors, amounts owed by distributors, which types of books are selling and where. A similar practice is followed by Belmont Tower Books, whose code consists of numerical designs not unlike those at the bottom of bank checks."

The article indicates that Dell adopted the barcoding system three years earlier, in 1971 or 1972.  This contradicts the reports of 1969 and 1970 paperback printings with the codes.  Either way, the system unquestionably predates the 1973-vested UPC.

There's still one mystery, though: What could be the meaning of the first bit in each 5-bit group? As far as the nine digits above are concerned, it's superfluous. If it were used for spacing or alignment, I would expect the values to be constant from position to position and from one title to another...but they're not. In five of the nine positions, the values are all over the place:

"Wasted" first bits of each 5-bit string
1 1 0 0 1 0 1 0 0
0 0 0 1 1 0 0 1 0
0 1 0 1 1 0 0 0 0
1 0 0 0 1 0 0 0 0

Maybe I'll also solve this someday. But until then...that's all.

17 April 2022



Yes, that's snow on the ground in late April.

No, there isn't much else I want to talk about.

©2002-22 Andrew Turnbull.

100% hand-coded from scratch XHTML!
This is a JavaScript-free website, and I'm proud of that.