For TM a date is a specific period of time connected to an event or text. Since time is (for our purposes at least) a linear progression, we have to choose a basic unit to measure it. We have opted for the day, as a lower granularity would enormously complicate counting without sufficient added benefit, given that in the ancient world only very few dates refer to smaller units than the day.
Working with traditional dating system with years, months and days is allright for humans, but for database purposes, we needed a simple number to refer to a unique day. Being blissfully unaware of the Julian day numbers used in astronomy, we have developed our own system with 1 January 333 BC as day 1, a serial number which increases by 1 for each day. For dates preceding 1 January 333 BC, we substract 1, starting with 0 for 31 December 334, and continuing with negative numbers for what precedes.
A period of time can thus be identified by a combination of two day numbers with an underscore ('_') in this system: www.trismegistos.org/time/450_450 thus refers to 25 March 332 BC, and www.trismegistos.org/time/135019_135383 (TM Time 135019_135383) to the period from 29 August 37 AD to 28 August 38 AD. The latter can in Egypt also be described as '(regnal) year 2 of Caligula', which we have also given a persistent identifier: www.trismegistos.org/period/538 or TM Period 538. For eponyms, we have for the time being still separate stable identifiers, e.g. TM Eponym 36 (www.trismegistos.org/eponym/36) for Caligula, although this may change in the future (we will make sure to put a redirect in place, of course). If there is a period (or eponym) identifier to your taste, it is better to use that as that is the chronological concept that humans generally refer to. In the case of centuries, for example, our 6th century AD is TM Period 3022, which currently corresponds to TM Time 303889_340413 (being 1 January 500 to 31 December 599). This 'alternative' conversion of a century goes back to the very earliest stages of the database, long preceding TM. We may well correct it to 1 Jan 501 - 31 Dec 600 in the future, but first want to wait for general scholarly conventions on how to deal with dates such as 'early second century AD' (TM Period 2720), for which about every single project has its own conversion into numbers.
The TM Time and TM Period identifiers refer to chronological concepts. This is also the level at which GODOT works, and we are implementing their stable identifiers now. There is, however, also another level at which we use persistent identifiers: that of the chronological expression as it is found in the ancient source itself. We have developed these in the course of the GODOT project and the top levels are already accessible in this version of TM Time, be it only through the 'Dates'-tab on the detail pages of TM Texts. An example is TM ChronNam 541 (www.trismegistos.org/chronnam/541) for the chronological concept '(regnal) year 6', independent from its application to a specific ruler. This entry in our CHRONNAM table is connected with several CHRONNAMVAR entries and in the level below that with CHRONNAMVARCASE entries. The latter levels are currently still inaccessible, but we hope to change that once we continue our work on TM Time, hopefully in the context of GODOT. Above that we are working on a CHRONNAMCOMB level, which will contain combinations of chronological entities, again purely on the formal level.
I am sure the last word has not been said about modelling dates, and we would be happy to discuss this with anyone interested.