I'm at the stage of formulating a methodology for comparison and have been messing around with GIMP. I'd like your feedback!
The first stage of my analysis will be to categorise reverse dies by their most easily discernible features. I consider this to be primarily the relative positioning and alignment of the date. Hocking describes two surviving reverse matrices of the sixpences in the RM museum, the first of which is dated "17--" only and the second of which is dated "178-" only (thanks @Rob). This suggests that the last digit or last two digits were hand-punched into the die, confirmed by the wide variation of their relative positions in my sample of about 500.
While the figures themselves degraded with die use, their relative positions should not have changed.
I intend to use GIMP to align the Irish shield as closely as possible to perfectly horizontal and vertical after which I'll resize the shield so that it becomes 300 pixels across. This will provide an absolute point of reference for each specimen in my database. A foreseeable drawback is that not all of my sample images will be of high enough quality to treat in this manner. It's possible, though, that they would have been unusable in any regard. Here's an experiment I did this morning:
The next stage will involve making each edited image semi-transparent and comparing the date positions two at a time, progressively developing categories which may then need to be further refined in order to determine dies.
I will also need to ensure that each of my samples is unique. My images have largely come from coins being sold. If they've been sold more than once in the last ten years or so it's possible that they appear more than once in my database. I can't have this as it will throw out my statistical description/inference of the relative number of coins produced by each die. A comparison of toning, mottling, haymarking, die degradation and coin wear should achieve this task.
I'm inviting criticism of my proposed methodology from anyone, especially those who've done die studies before.
Thanks!
Nathan