Why a New Classification System is Needed
When I started collecting star cancels, I relied on the different listings, each with their obvious benefits, but also many drawbacks. It quickly became apparent that there was a need for a new classification system which would satisfy the following requirements:
Completeness of the Listing
Listings serve as aids in determining the origin of a given stamp or cover. Unfortunately, only Proulx makes a serious attempt at completeness, and only for the state of New York, making its usefulness quite limited. With a new classification system, a serious attempt at completeness of the listing must be a top priority.
Detailed Dates of Use
Listings also aid in dating the use of a given stamp or cover. Most of the existent listings include some information regarding the dates of use, but generally this information lacks granularity, being limited to the years or the issues of use. A new classification system should seek to provide more precise dates of use by including the earliest and latest known usages.
Design Types
Precise enumeration of the different types of star cancellation designs is generally missing from the existent listings. Instead broad categories of designs such as shaded stars and six-pointed stars are provided. The result is that a given listing tells you little about the cancel's design. For example, consider the Skinner listings 'ST-6P 1' and 'ST-6P 2'. What is the difference between them? While the 'ST-6P' designation indicates they are six-pointed stars, without looking them up, one knowns nothing else about the design, such as if are they solid, outlined, or negative. A new classification system should clearly and completely define the design type for the listing.
While Proulx does enumerate the design types, it suffers from a different problem, namely, there is no relationship between one type and the next. As new design types are found, they simply get added to the end of the list. This results in types '10' and '35' being practically identical, but types '7' and '8' having virtually no similarities. A new classification system must address this by using designations for the different types which bear a relationship to the design, while simultaneously allowing new types to be added easily.
Groupings
One obvious problem with most of the existing listings is that not all of the star cancellations are even grouped together. For example, a star cancel which includes the word "PAID" might be grouped with the paid cancels instead of with the star cancels. Moreover, this problem makes it pain-stakingly difficult to enumerate all the star cancellations in existing listings. Like Proulx, this should not be a problem for the new classification system, as it only focuses on star cancellations.
One nice feature of the Skinner and Cole listings is that similar designs are grouped together. For example, all the shaded stars are generally found in a single section. However, similar to the problem above, because a given design might fit into more than one group, there is no way to know in which group a given design will be found. Thus, an important requirement for a new classification system is that designs can easily be grouped together. Additionally, a given design must be able to be easily placed in more than one group simultaneously.
Searching
Related to grouping is the issue of searching. Being able to easily find a particular star cancel to look up information about it is essential. Both the Skinner and Cole listings allow for searching by group, although as described above there are some problems. They also both have indexes to look up by origin of use, which works quite nicely. Proulx also allows for lookup by origin of use and takes it a step further by having indexes not only for each town, but also for each county. Additional search dimensions (such as date of use or ink color) could prove to be useful. And since a design should be allowed in more than one group, it follows that there is a need to be able to search by any characteristic used to define the groups.
A New Taxonomy of Star Cancellations
A new classification system that meets the requirements described above inevitably needs to be based on more than a single attribute. However, using too many attributes makes for a system that is unwieldy and difficult to use. Therefore, the choice of attributes is extremely important. Attributes can be fairly broad or extremely specific in scope. After much thought, four attributes were chosen given their capacity to describe fully most star cancel designs:
- Surround
- Describes any part of the design which is outside of the outermost star.
- Points
- Describes the number of points of the outermost star.
- Form
- Describes the appearance of the outermost star.
- Inside
- Describes any part of the design which is inside of the outermost star.
Each attribute can take on many values, and the choice of how to represent the different values is critical in creating a unique code for each distinct star cancel design. At first, I thought of enumerating the list of values for each attribute and then assigning a unique numerical code to each. The benefit of this approach is that it is fully extensible. As new designs are encountered, adding any new values to the attributes is simple and straightforward. However, to describe a star with the four attributes requires four numbers of variable length (in the number of digits), resulting in a code such as "11-2-1-18" or "11020118", neither of which are particularly easy to remember.
Instead of using a numerical code, I realized that I could use a letter code. This allows for enumerating the 25 most common values ('A' through 'Y') for each attribute, and one more ('Z') for any other value. The result is much simpler codes, such as "KBAR" (based on the same values as the numeric code given above). Because of the limitation to the 25 most common values, there is a need to annotate the code so as to describe any variations to the design. The annotations can refer to any of attributes, and not just for the value 'Z'. For example, a solid 5-pointed star with some negative text would have a code of 'ABAY'. If the text was the letter 'T' or the number '10', the annotated code would be 'ABAY { "T" }' or 'ABAY { "10" }' respectively. An added benefit to this approach is that it provides an additional mechanism for grouping.
Here are the four attributes and the codes for each:
| Code | Description | Code | Description |
|---|---|---|---|
| A | None | N | Non-touching outline square |
| B | Non-touching solid circle | O | Touching solid square |
| C | Touching solid circle | P | Non-touching outline pentagon |
| D | Non-touching outline circle | Q | Touching outline pentagon |
| E | Touching outline circle | R | Non-touching solid pentagon |
| F | Outline circle and non-touching solid circle | S | Touching solid pentagon |
| G | Outline circle and touching solid circle | T | Touching solid hexagon |
| H | Non-touching double outline circle | U | Non-touching outline hexagon |
| I | Touching double outline circle | V | Non-touching solid hexagon |
| J | Non-touching double outline diamond | W | Shooting star |
| K | Touching scalloped outline circle | X | Arc |
| L | Non-touching outline circle of dots | Y | Ellipse |
| M | --- | Z | Other |
| Code | Description | Code | Description |
|---|---|---|---|
| A | 4 points | N | 17 or more points |
| B | 5 points | O | Two 5-pointed stars |
| C | 6 points | P | --- |
| D | 7 points | Q | Four 5-pointed stars |
| E | 8 points | R | --- |
| F | 9 points | S | --- |
| G | 10 points | T | --- |
| H | 11 points | U | Two overlapping 4-pointed stars |
| I | 12 points | V | Two overlapping 5-pointed stars |
| J | 13 points | W | --- |
| K | 14 points | X | 5 points on 15 points |
| L | 15 points | Y | 5 points on 20 points |
| M | 16 points | Z | Other |
| Code | Description | Code | Description |
|---|---|---|---|
| A | Solid | N | Negative spokes |
| B | Negative solid | O | 'Mixed' solid-shaded |
| C | Cross-hatched | P | 'Pre' solid-shaded |
| D | Line-shaded | Q | 'Post' solid-shaded |
| E | Outer-outline | R | 'Pre' line-shaded |
| F | Negative outer-outline | S | --- |
| G | Outer-outline with spokes to arm points | T | 'Pre' line-shaded and 'post' solid-shaded |
| H | Outer-outline with spokes to arm joints | U | --- |
| I | Outer-outline with spokes to arm points and joints | V | Solid arms |
| J | Full-outline | W | Solid arms and solid center |
| K | Negative full-outline | X | Negative solid arms |
| L | Double full-outline | Y | Negative solid arms and negative solid center |
| M | Spokes | Z | Other |
| Code | Description | Code | Description |
|---|---|---|---|
| A | None | N | Negative non-touching outline circle |
| B | Non-touching outline circle | O | Negative touching outline circle |
| C | Touching outline circle | P | Negative non-touching solid circle |
| D | Non-touching solid circle | Q | Negative touching solid circle |
| E | Touching solid circle | R | Negative non-touching solid 5-pointed star |
| F | Non-touching solid 5-pointed star | S | Negative touching solid 5-pointed star |
| G | Touching solid 5-pointed star | T | Negative non-touching outer-outline 5-pointed star |
| H | Non-touching outer-outline 5-pointed star | U | Negative touching outer-outline 5-pointed star |
| I | Touching outer-outline 5-pointed star | V | Negative non-touching full-outline 5-pointed star |
| J | Non-touching full-outline 5-pointed star | W | Negative touching full-outline 5-pointed star |
| K | Touching pentagon | X | Negative cross |
| L | --- | Y | Negative text |
| M | Text | Z | Other |