How Many Are There?

A beautiful $1,000 bond of the Staten Island Rapid Transit Railroad Co. was available at the January, 2024 Dulles show. That certificate was issued and uncancelled and is currently cataloged as STA-891-B-40. When viewed in the online dataset, (at coxrail.com) you will find it ranked as F6 (Frequency 6.) Clicking on the blue hashtag near the catalog number allows you to see all serial numbers recorded. The database currently shows numbers 0000, 1022, 1023, and 1024. Until the Dulles show, all but one record of the bond had come from auctions by R.M. Smythe. The earliest records of the bond came from unillustrated lots in sales held 1991 and 1992. The first serial number recorded was 1024 which sold in 1993 and then again in 1997. Serial number 1022 sold in 1994 followed by a specimen in an unillustrated lot in 2004. Harmer sold a specimen with serial 0000 in 2007. I never encountered any other appearances until the newest serial number 1023.

So how rare is a certificate variety that does not seem to appear for sale for seventeen years? Or twenty-seven years when talking about an issued certificate! In my parlance, I'd quantify it as, "pretty darn rare." My ranking system, however, does NOT quantify "rarity." Rather, it calculates the average time interval between appearances at the sub-variety level. The assumption is that the frequency of appearances in the past will probably repeat in the future. Appearances may be either offerings or sales. My system also tallies reports from correspondents because each one represents a purchase opportunity at some point in the past. It is important to understand that frequency of appearance calculations have nothing to do with serial numbers. Reports of serial numbers in auctions, price lists and from dealers and collectors have always been minimal. Currently, only 15.4% of all reports have been accompanied by serial numbers. (160,268 serial numbers versus 1,038,618 records of appearances.) While I report every serial number I can find, the actual numbers of items in existence are normally higher and, in some cases, dramatically higher.

For example, Smythe was a terrific source of varietal price information about North American railroad certificates. All told, Smythe catalogs and related materials contributed 790,151 records of offerings. Of that number, I was able to acquire only 32,914 serial numbers, most of which entailed close examinations of halftone photos with a hand lens. My long-term goal has been to acquire as many serial numbers as possible from every collector, catalog, auction, mail bid, price list, book, magazine, and other sources depending on available time. For accurate price records, I generally record only single-item lots. With more time, I could pick up many more serial numbers by tearing apart multi-item lots.

Getting back to the specific case of the Staten Island Rapid Transit Railroad bond, I have learned about a total of four specific serial numbers from eight separate offerings. I use a mid-1980 start date for my project in order to include all of the early references to railroad certificates such as LaBarre, Narbeth, Hendy, Hall, and a handful of dealer price lists. That was 43.6 years ago. Every one of those eight meager appearances represented a legitimate opportunity for someone to have purchased an example of STA-891-B-40. On average, there was one purchase opportunity every 5.44 years. If we consider only issued certificates, there would have been one opportunity every 8.7 years. If we stick with the four serial numbers we can prove conclusively, opportunities decrease to once every 10.9 years.

In truth, no quantifications can be fully accurate because we have no method of learning about additional certificates that might have been offered or sold out of public sight. Similarly, collectors cannot possibly learn about every offering in the future either. Since both collectors and I are working with incomplete information, calculated frequencies of appearance should be reasonably valid for both of us.

Therefore, I stress that a calculated "frequency of appearance" system is merely an approximation of how frequently "sub-varieties" have appeared in the past, and therefore likely to appear in the future.

It should be clear that frequency-of-appearance estimates are "flexible." A collector might be able to shorten a waiting period by committing time to find a desired certificate in a dealer inventory somewhere. We all know there are going to be new discoveries and inevitable losses.

There has been a proliferation of collector exits from the hobby in the last few years. Many certificates appeared after being held in "strong hands" for years, sometimes decades. Recent releases have made many scarce and rare certificates appear more common than they really are.

The reverse is also at work because several big-time collectors have donated certificates to "lock `em up" repositories. Those donations have effectively removed unknown numbers of rarities from future acquisition. For example, none of Bob Greenawalt's extensive railroad collection made it into my database because he did not share collection contents with me while he was alive. He gave his collection to the Huntington Library which has neither published images nor even an inventory. We know almost nothing about many extreme rarities he had collected.

Before I decided on my "frequency of appearance" system, I had examined twelve or thirteen rarity systems designed by giants in several hobbies. ALL those systems were based on EXPERIENCES of those experts. I was highly disinclined to ever claim such status. I was especially reluctant to estimate the number of certificates held by dealers and collectors. Moreover, I was never able to understand how "rarity estimates" could tell collectors how long they might need to wait for purchase opportunities to come around. I asked, "Why not just let the rate of past appearances predict the future and remove opinions from the mix?"

In summary, recorded serial numbers say little about rarity. By themselves, serial numbers are merely a tracking system. They say much more about my ability to find images and convince collectors and dealers to contribute information. With rare exception, serial numbers are NOT a measure of the number of certificates known, but rather the minimum number that might exist.