Purpose. The ultimate reason to record serial numbers is to help us get a "feel" for the rarity of different types of certificates.

But let's not overstate the case. Rarity does NOT establish value. By itself, rarity influences desirability and hence influences value to some degree. Yes, it is true that all other things being equal (age, attractiveness, condition, autographs, etc.) rare certificates often cost more than common certificates. However, there are thousands upon thousands of examples where common certificates are worth more than rare certificates. Do not confuse yourself into thinking that rarity is the most important influence on value.

It would be nice...to have a full record of all serial numbers from every variety of certificate. That will never happen. The most we can ask is a beginning.

By mid-2011, about 103,000 serial numbers have been recorded, which means we are only at the beginning. The more serial numbers we record, the better we can understand which certificates are rare, and which are common.

The secondary reason to record serial numbers is to help discover new varieties. Gaps or differences in serial numbers often tell us to look more closely at certificates that flank the gaps. So far, the study of serial numbers has disclosed a couple hundred new varieties. More are waiting to be discovered.

I use a $25 minimum when recording serial numbers from eBay sales. Otherwise, I would not have time for anything else in my life. However, that means I am missing large numbers of serial numbers.

NEVER assume that value is related to the total number of serial numbers reported. Common certificates are greatly under-represented in the database.

I REQUIRE images of certificates for all serial numbers. I generally ask to see certificates before I enter any new serial numbers into the database. I do this for several reasons:

• It prevents the insertion of fictitious numbers.
• It allows review of images when trying to confirm new varieties.
• It allows review in case serial numbers seem out of date order.
• Even the most picky collectors make mistakes in identifying certificate varieties.
• Even the most picky collectors make mistakes in deciphering handwritten dates and serial numbers.
• Numerous tests have proven that it takes longer to enter info from lists of serial numbers than from copies.

"Some of your serial numbers seem wrong." Agreed. Some serial numbers in the database are wrong. Handwritten serial numbers and dates can be very hard to read. Mistakes are common. Even the most experienced dealers and collectors make mistakes.

Serial numbers come from thousands of sources and every source is prone to error. Be assured that accuracy ranges from perfect to highly questionable.

Sources of serial numbers in the database:

• copies of certificates sent by collectors and dealers
• lists of certificates in museum and private collections
• serial numbers listed in sale catalogs
• photos in auction catalogs
• photos in books
• photos in dealer price lists
• photos on eBay and other internet auction sites
• photos on dealer internet sites

I do NOT include serial numbers from bond coupons. Just because coupons exist does NOT mean collectible bonds exist.

Mistakes occur when trying to read characters of similar shapes, especially when characters are handwritten. These groups of letters and numbers create the greatest number of mistakes:

• O, 0, U, D, C, G
• B, E, F, P, R
• 1, 4
• 2, 7, 9
• 3, 8
• 6, 0

Prefix and suffix letters. While serial numbers often carry prefix and suffix letters, catalogers rarely report such characters. As a rule, catalogers also drop leading zeros. This is too bad, because prefix, letters, suffix letters and leading zeros are very important.

Oftentimes, prefix letters indicate specific series of certificates. For instance, "C", "V" and "M" were commonly used as prefixes for serial numbers on $100, $500, and $1000 bonds. If a "C" serial number on a bond is reported for a variety otherwise populated by "M" prefixes, then there is definitely an error.

Serial number prefixes were also often used to distinguish between common ("C") and preferred ("P") stock certificates.

Less frequently, companies used "C", "P", "B", and "N" prefixes to indicate certificates were issued in Chicago, Philadelphia, Boston, or New York. This images indicates a serial number for common shares issued in New York.

Prefix letters on late-date certificates often indicated different series of issuances. This is particularly true as heavy trading in the 1960s and 1970s prompted the issuance and redemption of tens of thousands of certificates per day.

If you find one or two serial numbers out of date order in a large sequence, then chances are very great that dealers and collectors misread dates. Sometimes certificates carry two and sometimes three dates and multiple date can help clarify problematic numbers.

However, there are hundreds of examples of valid serial numbers that were definitely issued out of date order.

How can date order could get messed up? It may seem strange, but it is quite common to see low-numbered certificates issued out of date order. Sometimes, low-numbered certificates were dated several months after companies incorporated and after higher-number certificates had been issued. It seems possible that such certificates might have been set aside for "special" investors and friends of management.

Companies often issued temporary certificates immediately after incorporation and replaced them a few months later with engraved "permanent" certificates. Depending on the company, permanent certificates often replaced identically-numbered temporary certificates. Some companies issued their replacement certificates with original dates and some used actual dates.

Normally, when stock certificates were sold to new investors, old certificates were sent back to companies, which then transferred ownership and issued new, higher-numbered certificates. For unknown reasons, a few companies reused their old numbers. When issuing new certificates, a few companies indicated their actions with suffixes. Sadly. most catalogers fail to report suffix letters.

Companies often used prefix letters to indicate cities where certificates were sold. For instance, certificates C1 to C3000 might have been issued in Chicago and certificates N1 to N27000 might have been issued in New York. Certificates issued from different cities never sold at the same rates. Therefore, when collectors and dealers fail to report prefixes, serial numbers and dates will seem highly confused.

Similarly, batches of certificates may have been divided between two or more brokers, trust companies, transfer agencies, or exchanges. In those cases, serial numbers would NEVER have been issued in strict date order.

Gaps in numbering. Companies often switched designs of certificates as they became successful. As a rule, most companies maintained their original numbering schemes, but often jumped a few hundred or a few thousand numbers between successive certificate orders. Consequently, large and legitimate gaps may occur in serial numbering.

Unless entire books of certificates have been reported, gaps occur in every variety. Take, for instance, BAL-662b-S-30. About 180 numbers of extant certificates have been reported out of a known range of 58,226. Do the other 58,046 certificates exist? Maybe 10,000 do. I doubt, however, today's collectors will ever report more than 200 more. We will be left with thousands of gaps in numbers. Regardless of how many certificates we may think exist, most of us would agree that S-30 from the BAL-662b, Baltimore & Ohio Rail-Road Co. is a very common certificate.

On the other hand, collectors have only reported six examples of stock certificate variety S-37 from a later name variant of the B&O (BAL-662d). These six meager examples span a 600 number range, so there are large and significant gaps in numbering. How many of those certificates exist. Ten? Twenty? Fifty? Who knows? The gaps in the numbering of BAL-662d-S-37 are much more significant and meaningful than the gaps among BAL-662b-S-30.

Repeated numbers. Unfortunately, each time companies changed certificate designs, some re-started their numbering with certificate #1. Repeated serial numbers are fairly common. Sometimes they signal major corporate changes. Sometimes they mean absolutely nothing.

You WILL find strange numbering. It is best to assume that all examples of strange serial numbering had legitimate reasons. However, explanations for such strange corporate behavior are now lost to history and you will drive yourself crazy trying to discover the reasons.

Some strange numbering results from listing behavior by catalogers and dealers. There are many instances where auction catalogs (particularly German catalogs) show pictures of certificates not actually being offered for sale. If correspondents send such images to me, they may not realize they are mismatching dates and serial numbers.

These inadvertent creation of the kinds of mistakes is the main reason I now require images before adding serial numbers to the database.

Low serial numbers. European collectors frequently pay more for certificates with very low serial numbers. So far, low serial numbers seldom command much of a premium among U.S. collectors.

How "low" is "low"? European collectors often pay 25% to 40% more for #1 certificates and perhaps a 10% to 20% more for #2 certificates. #3 certificates sometimes command a 10% premium, but certificates numbered above that level rarely attract attention.

Do not be too overly demanding in your pursuit for low serial numbers. They may not exist.

Remember that railroad companies often continued their old serial numbering when they switched certificate designs. Therefore, the lowest serial number for a particular stock certificate designs might have been 101, 1,001, or even 100,001.

In fact, a recent review of reported serial numbers strongly suggests that over half of all varieties of stock certificates never had serial numbers below 100. Simply put, you have no hope of finding serial #1 for over half of all varieties of stock certificates because they never existed!

Need more convincing? A deeper review of the database suggests that almost one third of all varieties of stock certificates had a minimum number above 500.