This article looks at the types of repertoire included in eighteenth-century London concerts. As discussed in the first article of this series, information on the works performed is encoded, in a complicated way, in the “programme” field of the dataset.
The data is based on concert advertisements in newspapers, so there is considerable variation in the detail provided. Some advertisements spell out details of all of the works and who will perform them, but it is more typical for the focus to be on the performers, with the works often vaguely specified, such as “a concerto by Handel” (if you are lucky, it will say what instrument it is for).
The “programme” field for each concert consists of several entries – separated by semicolons – one for each item on the programme. Each entry includes information in a standard order – Composer Genre Instrument Performer – with the “Genre” and “Instrument” indicated by standard abbreviations (which are listed in the explanatory note that accompanies the dataset). So the data can be parsed by splitting it at the semicolons, searching each entry for the standard abbreviations, and using the order to deduce the identities of the composers and performers (considered in a later article).1
Let’s start with a chart showing the proportion of concerts that included different genres. These are just the top ten genres, broken down by decade:
Over 25% of concerts in the 1750s and 1760s included an Oratorio (“O”), but these declined in popularity (Handel died in 1759) in favour of Concertos (“CN”). Solos (“SL”) declined substantially from the 1780s, whereas Glees (“GL”), Overtures (“OV”) and Symphonies (“SYM”) increased around the same time. Songs (“SG”) remained popular throughout.
The following chart (again just showing the top ten genres), shows the average number of pieces in each concert including that genre. So, in concerts that included a song, there were, on average, four songs. In concerts with a glee,2 there were 2½ of them, and concertos were often performed in pairs.
To simplify the analysis of the multitude of genres, I have derived a couple of other variables.
GenreVocal is TRUE for concerts that include any genres with singers, and FALSE for purely instrumental concerts.
GenreSize reflects the forces required, being Large (more than six players), Medium (more than two players) or Small. This is an approximate classification, categorized as follows:
- Large: concerto, concerto grosso, oratorio (ode, opera etc), overture, sinfonia concertante, symphony
- Medium: anthem, catch, chorus, divertimento, glee, quartet, quintet, sextet, trio
- Small: duet, favourite, lesson, notturno, piece, recitative, serenata, solo, sonata, song, variations
The distribution of concerts (excluding the 25% or so with no genre information) according to these categories is as follows:
Over 86% of concerts included at least one singer. Even in otherwise instrumental concerts, there were often a few songs between larger pieces. More than 91% of concerts in the dataset were “Large” – maybe not by today’s standards, but they would have required a reasonable amount of space and perhaps a conductor/director of some sort. Of course, this is a dataset of advertisements, and the promoters of larger concerts may have been more likely than those organising smaller affairs to need to, or be able to, put advertisements in the daily press. Small concerts might have been advertised in other ways.
We can analyse these categories against other variables such as venue type, ticket price, etc. There are lots of possible combinations, so I won’t go through them all, but some of the more interesting results are as follows:
- Houses were most likely to host Medium concerts (9% of House concerts were Medium, compared to 2.6% overall), and Taverns most likely to host Small concerts (15% vs 5.7%). Almost all Church and Theatre concerts were Large (99% and 97% respectively).
- Instrumental concerts were most likely in Taverns (28%, compared to 14% overall), Houses (21%) and Halls (17%).
- “Miscellaneous” and “Garden Benefit” concerts were most likely to be Small (49% and 19% respectively). The Miscellaneous and “Concert Benefit” types were also most likely to be instrumental (26%).
- Larger concerts tended to be more expensive (apart from those in Theatres – see previous article).
- Daytime concerts were more likely to be Small (18% of Daytime concerts) or Instrumental (19%).
- As time went on, there were fewer advertisements for smaller concerts. Over 10% of advertisements in the 1750s and 1760s were for Small concerts, but this fell to just 1.3% in the 1790s. This might be because there were fewer Small concerts, or perhaps they became less likely to be advertised in the press.
- Instrumental concerts comprised about 19% of the total in the 1750s-1770s, but this fell to just 5% in the 1780s, recovering to 12% the following decade.
We can also look at the combinations of genres that tend to (or not to) be programmed together. To do this, we restrict our attention to concerts containing at least two genres, and, for each combination of genres, compare the actual number of concerts with how many would be expected if genres were distributed at random in the same proportions as in the actual data. As well as the ratio between the actual and expected numbers, we also need to consider the statistical significance – expecting 100 and seeing 200 is much more significant than expecting 1 and seeing 2. One simple approach is to ignore any combinations where the actual and expected numbers are too small (perhaps less than 20), or by considering other statistics – such as the (O-E)2/E statistic that is the basis of the Chi-squared test (where O and E stand for the Observed and Expected values respectively).
In this case, we find, for example, that Catches and Glees have a particular affinity, as do Duets and Trios, Overtures and Quintets, and Quartets and Sonatas. The combinations that tend to be avoided include Glees and Solos, Oratorios and Sonatas, and several other pairs at opposite ends of the size spectrum.
This is quite messy data, and can be analysed in many different ways. Such data exploration can often reveal questions for further investigation. In this case, for example, we could look more closely at the reasons for the changes in the relative fortunes of different genres over time – especially the marked shift that seems to have happened in the 1780s.