As the number of TEI encoded titles in Digital Lacy is about to reach a milestone, it seems a good moment to consider its representativenessity. So I have been dusting off my utterly decayed spreadsheet knowledge, and doing some counting. I can compare (fairly easily) the following characteristics for all plays in the LAE:
– first performance date
– size
– genre
Other possibly useful categories might include
– popularity as measured by the number of performances or revivals
– text type (translation, adaptation, burlesque etc.)
– resource usage (size of cast, effects, music etc.)
Reliable and consistent data covering these is however rather more difficult to compile.
First performance date
This table shows the number of titles claiming a first performance date for each decade between 1820 and 1880 in each of the complete LAE, the subset of it included in the VPP, and the subset of it available in TEI format. It shows that there are 1512 titles in the LAE, of which only one was first performed in the 1880s, a decade barely represented in the LAE. Moving back in time, however, 99 titles (or 6.6% of LAE titles) have a premiere dating from the 1870s. The table allows us to compare this with the equivalenmt figures for VPP (which has a total of 335 titles) and TEI (total so far of 99). We can see that the TEI sample (at 5%) is close to the Lacy proportion, while the VPP sample (2.3%) really should have included more titles from this decade.
| total | 1880s | 1870s | 1860s | 1850s | 1840s | 1830s | |
| LAE | 1512 | 1 | 99 (6.6%) | 442 (29%) | 406(27%) | 186(12%) | 164(11%) |
| VPP | 335 | 1 | 8 (2.3%) | 115 (34%%) | 122(36%) | 43(13%) | 28(8.3%) |
| TEI | 99 | 0 | 5% | 20% | 35% | 13% | 12% |
For each decade, except the 1860s, the TEI proportion is close to the proportion of titles found in the whole Lacy corpus, which is encouraging. It does appear however that we need to add one or two more titles from the 1860s – and maybe remove some from the 1850s.
For titles premiered up to and including the 1820s, the situation is slightly different:
| 1820s | 1810s | 1800s | 18th c. |
| 91(6%) | 32(2%) | 16(1%) | 15 (1%) |
| 11(3%) | 2(0%) | 0 (0%) | 3 (0%) |
| 2 | 2 | 0 | 4 |
These figures suggest that VPP has made a determined effort to exclude older titles; the TEI however seems to over-represent titles from the 18th century, some of which might perhaps be removed.
Size
As an approximation for size, we use the number of pages in the printed edition, allocating each title a value of Large if it has more than 45 pages, Medium (25-44 pages), or Small (fewer than 25 pages).
Calculated in much the same way as above, the following table shows some discrepancies in the the TEI sample:
| total | large | medium | small | |
| LAE | 1512 | 209 (13.8%) | 547(36.1%) | 755 (49.9%) |
| VPP | 335 | 50 (14.9%) | 125 (37.3%) | 160 (47.7%) |
| TEI | 99 | 14 | 26 | 59 |
The TEI sample has too many small texts, and not enough medium sized ones. (I admit it: the short texts are generally funnier and easier to encode rapidly)
Genre
Deciding on a genre for each play is a tricky matter, which I explored in this blog some time ago (see https://foxglove.hypotheses.org/828 and https://foxglove.hypotheses.org/876 ) Amongst other problems, the terminology used by the theatre (and hence the plays) is not stable. It’s clear that broad categories such as “drama”, “comedy” etc. fragment into particular niche terms (“melodrama”, “equestrian drama”, “burlesque”, “burletta” etc.) which then go in and out of fashion. A classification derived from formal properties (presence or absence of songs, use of verse, presence or absence of traditional characters, etc.) might well be preferable but is hard to implement (there’s a boot strapping issue here). So I have fallen back (again) on my own simplistic 15-category taxonomy. Which produces a table like this:
| Total | FARCE | DRAMA | COMEDY | SPECTACLE | BURLESQUE | |
| LAE | 1512 | 461 (30.49%) | 339 (22.42%) | 190 (12.57%) | 121 (8.0%) | 86 (5.69%) |
| VPP | 335 | 93 (27.76%) | 109 (32.54%) | 25 (7.46%) | 31 (9.25%) | 27 (8.06%) |
| TEI | 99 | 36 | 21 | 11 | 7 | 5 |
| MELODRAMA | COMEDIETTA | BURLETTA | OPERA | TRAGEDY | PANTO | BALLET | |
| 75 (4.96%) | 75 (4.96%) | 50 (3.31%) | 47 (3.11%) | 38 (2.51%) | 11 (.73%) | 2 (.13%) | |
| 17 (5.07%) | 11 (3.28%) | 10 (2.99%) | 1 (.3%) | 4 (1.19%) | 3 (.9%) | 0 (.0%) | |
| 6 | 2 | 4 | 1 | 4 | 1 | 0 |
This, to my surprise, shows that the proportion of TEI titles to LAE titles is more or less identical (i.e. they differ by about 1%) for all but a few categories, notably (and unsurprisingly) FARCE which is over-represented in TEI, as is (unexpectedly) TRAGEDY, while COMEDIETTA is mystifyingly under-represented.
Conclusion
The astute reader may have been fretting somewhat that I have presented as percentages values based on a total of only 99, rather than 100. This is because I am about to add the 100th title – and will of course be making sure that this one, at least, contributes to a balancing of the corpus. On the basis of the foregoing, I will need to select a title which is a COMEDIETTA, of medium length, and was first seen on stage during the 1860s. A quick Xpath shows me that there are 17 candidates: I narrow this down to six by adding the requirement that the title also appears in the VPP subset.
As we say in Xpath `div[@type='work' and ends-with(@ana,'COMEDIETTA') and starts-with(@ana,'M') and bibl[@type='printSource']/idno[@type='vpp']] … and then start browsing to choose one by an author as yet unrepresented, or with an interesting subtitle, as usual.


