Thursday, January 17, 2013

Measuring the Diversity of Immigration using the Old Bailey Online 1674-1834

"Mother's Wartime Passport -1941" A. Davey
This is the second in my series of posts on the Old Bailey Online (OBO) corpus. I've downloaded all of the trial transcripts from 1674 to 1834 (find out how on the Programming Historian 2), which is about 100,000 trials and 51 million words of text. In the last post I looked at the impact of editors and scribes on the vocabulary in the Old Bailey Proceedings.

This time I thought I'd look at something a little closer to my area of expertise: immigration to London in the Early Modern era. I've used the OBO heavily in my doctoral work on Irish immigrants, but that's been focused exclusively on the years 1801-1820, immediately following the 1801 Union of Irish and British parliaments. I've yet to take a longer look at immigrants across the centuries using the OBO and I thought this would be the perfect opportunity to do so.

This time I'll be looking at the "people words" extracted from the OBO corpus. As I mentioned in the last post they were identified by extracting all of the words that appeared between a set of "persName" tags in the XML version of the transcripts. This gave me just shy of 62,000 unique strings (referred to hereafter as "words") used to represent people. That's nearly half of all unique words in the corpus. Of those 62,000 words, most (55,000) are not found in the four English language dictionaries I used to identify English words. The remaining 7,000 are words such as "green" or "woman" or "the", which are used to refer to people such as "the woman" or "John Green", but which can also be used in other contexts (the woman's green hat). Not all of these words are therefore proper names; instead, they are words that have been marked up by the OBO team as a reference to a person somewhere in the corpus.

In Figure 1 you can see the rate at which these new "person words" appeared in the corpus.

Figure 1: Total number of "person words" found in the OBO corpus to date. [expand +]

Nothing particularly exciting here. It seems like most of the person names that are also English words appear very early on. It also looks like the number of unique words used to describe people increases at a steady pace throughout the long eighteenth century. From a cursory look at the list of names, it seems evident that many of these words are surnames.

Given names (first names) on the other hand, are much less common. That's because most early modern Londoners had pretty common given names (William, John, Elizabeth, or some variation thereof). Silly names for babies are largely an invention of twenty-first century Hollywood actors and football players.

While London is home to hundreds of thousands of people in the eighteenth century, it's safe to say the number of surnames people had in London increased over time as migrants flooded in from across England and beyond carrying new names with them. New surnames therefore have a few ways to end up in the corpus:
  1. An established London family is mentioned in the record for the first time
  2. An immigrant family with a new name arrives in the area and ends up in the record
  3. Someone with a funny accent tries to say their name and it gets spelled phonetically
In the case of #1, it's entirely possible an established family (or anyone with that name) just avoided the Old Bailey for decades on end. I've managed to do so and there's no reason to expect others didn't too. However, common names shared by many people and local to Londoners should eventually show up. In fact, there's a reasonable chance they'll show up very early. We see this is in fact the case, as Smith, Wilson, Brown, White, and Allen all appear for the first time before 1680. "McCaffrey" on the other hand doesn't show up until 1834 and it's safe to say represents a name brought to London by an immigrant (either scenario #2 or #3 above).

New names arriving in the area may not be indicitive of the total number of new people who have migrated to London, but I do believe it reflects the growing diversity of immigrants arriving. Malcolm Smith and Donald MacRaild's article, "The Origins of the Irish in Northern England" shows that at least with Irish surnames, most names can be pinpointed to a particular region in Ireland. This regionality of names was still evident into the middle of the nineteenth century and will be no surprise to any genealogist who has sought out their family's past. We can see direct evidence of this regionality by mapping surnames. Great Britain Family Names allows you to see the distribution of any name in Britain in 1881. In Figure 2 you can see the distribution of "Howard" families, which clearly cluster around a few areas.

Figure 2: The distribution of "Howard" families in 1881 [expand +]
John Mannion agrees with Smith and MacRaild's conclusions about migration. In "Old World Antecedants, New World Adaptations" he argues that people tend to follow migration "channels". That is, someone from their village went before them and came home to say how great it was. Mannion was looking specifically at Irish migrants to Newfoundland and was able to show that villages who had already sent migrants to Newfoundland were vastly more likely to continue to do so than somewhere without the same history. That means the first "McCaffrey" in London was far more adventurous than the 351st. In fact, we could suggest that in many cases the first McCaffrey drew the other 350 over time by breaking the ice. I am interested in why that first McCaffrey decided to come to London, and what factors might have influenced his or her decision to do so.

The reason I think the OBO corpus is a useful set of records for monitoring this growing diversity of migrants is because I'm quite firmly convinced that the Old Bailey is the place one takes people they don't know when they've wronged you. I believe strangers (including immigrants and migrants) were much more likely to be subjected to the official justice system than were people with deep roots in the community.If a stranger wronged you, they had to be caught and punished quickly, or they might disappear. If your long-time neighbour steals your linen tablecloth, you have lots of options for how best to deal with them. You could smack him, you could set the dogs on him, you could knock on the door and demand it back. You had options, and time, because you knew he would be there tomorrow. And the next day. You don’t have that same assurance with someone you've never seen before. And that meant I believe people were more likely to seek a legal response than to choose a community resolution when dealing with strangers or those they do not know very well. In John Beattie's wonderful book, Crime and the Courts in England: 1660-1800, he appears to agree:
In the small-scale society of the village a prosecution may not have been the most effective way to deal with petty violence and theft. Demanding an apology and a promise not to repeat the offense, perhaps with some monetary or other satisfaction, may have been a more natural as well as a more effective response to such an offense, or perhaps simple revenge directly taken (Princeton: Princeton University Press, 1986, p. 8).
If I am correct in my assumption then immigrants are more likely to appear in the Old Bailey records than established members of the community (at least as a defendant), and are even more likely to do so shortly after they arrive in London as opposed to several generations later. That means that there is likely a reasonably strong connection between the date a name first appears in the OBO corpus and the date that name first appeared in the London area. It may not be a precise way to measure the arrival of new names, but I would hasard to say that in most cases it's probably accurate to within a few years or a decade at the most.

Therefore, one way to find new families with few if any connections to the locals arriving in the area is to look for the first time a given surname appears in the records. Considering the nature of the Proceedings, most names that appear in the record likely refer to people in London as opposed to strangers living far away. That's not always going to be true but for the most part it's a reasonable assumption. That means by measuring the rate at which new names appear in the OBO, we should get a reasonable if rough idea of the rate of immigration from distinct family groups over time.

To isolate surnames I've taken all of the 60,000 names present in the London area in the 1841 census and checked them against the "person words". This returned a list of just over 20,000 surnames. That means one third of all unique surnames in London at the end of the period have showed up in the Old Bailey corpus at some point or another. I'm sure I've missed a few, particularly those spelled phonetically, but this is probably a pretty good start.

You can see when each of these names first appears in the corpus in Figure 3.

Figure 3: Number of new surnames in the OBO corpus per year [expand+].
What Figure 3 shows is that the rate of new surnames arriving is actually fairly stable over the course of the eighteenth century. As mentioned in the previous post, the big dip around 1700 is caused by missing data and very short trial accounts, and we might be wise to assume that the entries around 1715 should actually be lower if we had the full set of trials as more names would appear earlier, filling in the gap. The long slow decline therefore over the course of the eighteenth century might actually be better understood as a reasonably flat line hovering around 100 new surnames per year and declining slightly towards 50 or 60. But it turns out that is not the whole story, and the clue to that is the increase in new names in the years immediately following the Napoleonic Wars c. 1815.

After the fall of Napoleon at Waterloo it seems quite clear that there's an influx of new surnames into the London area. I've got a suspicion that the cause of this influx is decommissioned soldiers and sailors who were dumped in London (or found their way there) after the war and got themselves into trouble. War collects soldiers and sailors from far and wide and brings them together. When those soldiers and sailors are no longer needed they're released to go on with their lives.

It would seem that after two decades of war enough people had been uprooted from their native regions by this process for a long enough period that they felt no inclination to go back home. Instead some of them obviously resettled in London or the growing industrial cities in the north, which seemingly offered greater opportunities or were more germane to their skills than the family farm. In fact, many people may have found themselves without a farm to go back to, since the enclosure movement had been consolidating ariable land into much larger units throughout the second half of the eighteenth century, leaving many people landless. That landlessness may have attracted them to the army or navy in the first place, and now with military life behind them they had to find something else to do with themselves. London, it would seem, was it. At least for some.

This trend of more new names after the Napoleonic War doesn't appear to be an isolated incident; it's merely the most obvious case. Instead we see similar patterns in other major wars and domestic conflicts involving the British, the results of which can be seen in Figure 4.

Figure 4: Number of new surnames in the OBO corpus per year, colour coded to highlight periods of war and peace [expand+]
Figure 4 shows the same number of new surnames per year arriving in the OBO corpus, but this time is highlighted to show some of Britain's major wars and domestic conflicts in the latter-half of the eighteenth century. The wars depicted in red are:
  1. The Seven Years War (1756-1763)
  2. The American Revolutionary War (1775-1782)
  3. The French Revolutionary Wars (1793-1802)
  4. The Napoleonic Wars (1803-1815)
While the American Revolution wasn't officially settled until 1783, it was effectively over by the end of 1782. The grey bar between 1802 and 1803 separates visually the two wars with the French.

The black bars represent years in which significant domestic conflicts took place:
  1. The Jacobite Rising of 1745 (1745-1746)
  2. The Gordon Riots (1780)
  3. The Irish Rebellion (1798)
In nearly all cases we see a decrease in the number of new names showing up shortly after a war or domestic conflict erupts. This is most evident for the Jacobite Rising of 1745. The apparent dip in migration at this point makes sense; who wants to move when there's a rebellion going on? This dip is then followed by lower than average numbers of new names until the conflict ends, at which point almost invariably the following years experience an above average result. This is evident both for domestic conflicts as well as international wars. The pattern appears again and again.

The differences between the average number of new surnames per year during war compared to the average in the five years after the end of a war are in fact statistically significant, at least for the American Revolution and the combined French Wars (paried t-test: p = 0.0418, and p = 0.0001 respectively. Significance in this case was p < 0.05). The Seven Years War does not pass the statistical t-test (p = 0.1346), However, I am confident we are seeing evidence of the same trend. While my statistical skills are rather rudimentary, I think it's worth noting that failing a t-test does not mean something is not true. Instead, it suggests the numbers alone cannot support that conclusion beyond all reasonable doubt. For me, the fact that the latter wars are so obviously following this trend strenghtens my confidence in a similar trend for the Seven Years War and we can see this in Figure 5, which shows the average number of new surnames per year during and after the three wars.

Figure 5: The average number of new surnames per year during the wars and in the five years following the wars [enlarge+]

The strength of the correlation between these conflicts and the decrease in new names, followed by an increase in peacetime suggests to me that my original assumptions about newcomers getting in trouble with the law were correct. It also suggests some interesting things about migration patterns in the eighteenth century. That is, people migrated when they felt it was safe. During times of turmoil, they stayed put and waited things out.

There are implicitly two groups of people here, so each requires its own discussion, I think. Firstly there are the sailors and soldiers. The reason we don't see these people arriving in London during wartime is perhaps obvious: they were in the employ of the state, off fighting the enemy. Gathered from across Britain and Ireland, as mentioned above, when they were decommissioned they had the opportunity to move where they liked and it would seem some chose London. This may have disproportionately included sailors who may have hoped to find work in London's booming shipping business.

The second group are families or individuals who have decided for economic reasons to move to London. Since we don't have evidence that someone with that name lived in London previously, many of them are presumably amongst the first of their stock to try their hand at London living. This in itself should not be taken lightly, as moving to early modern London without a social support network was an incredibly lonely and dangerous prospect, which is why so many migrants failed and found themselves in gaol, or starving and desperate, looking for any chance to get away. Sadly, we see many immigrants like Sarah Holmes, who claim that they "have no friend but God" as they throw themselves at the mercy of the courts.

What does all this mean? What can we learn about these arriving surnames? War and domestic conflict are not the only variables at play here, but I think it puts forth a reasonable case for the effects of war and peace on migration patterns of those moving towards London in the long eighteenth century. Returning to the idea mentioned earlier about the first McCaffrey (or the first of any family), it seems that families were only too willing to bide their time during periods of war, waiting instead for peace before making their way to a new life in London. We can see why this strategy might have been appealing. Why take a risk when the country is at war?

Unfortunately it may have been the wrong approach from an economic standpoint. According to Ball and Sunderland's "An Economic History of London, 1800-1914", the gap between real wages and cost of living peaked just after peace was called with France in 1817 (p. 95). That means people were most desperate when the government realised it actually had to start paying for the war it had waged. It seems to me likely that the two trends are actually connected. As new families arrived they may have been desperate for any work, forcing down the price of labour in London as a surplus of workers vied for jobs. It may seem counterintuitive, but these data suggest it's best to move during war rather than after.

Nevertheless, this unlikely set of criminal records has provided, I think, an interesting window into wider migration strategies across the eighteenth century. And it came about not by looking at how many people arrived, but when unique groups of people likely first emerged. London it would seem was home to an increasingly diverse population. That population continues to diversify to this day. And though I'm sure there are some Brits who might see war as a strategy for keeping the net migration in the "tens of thousands", take heed, for when peace comes, so too will the immigrants.

2 comments:

Tim Hitchcock said...

Great post - I wondered, however, if you had factored in the transition from 1789, after which the OBP were directed to reflect trials fully and accurately. There is also the censorship 1790-92. My impression is that lots of repetitive witness statements were excluded prior to the 1780s, the subsequent inclusion of which might contribute to the transition you have identified.

Adam Crymble said...

Thanks for your comment Tim. I havn't factored any changes in the sources - nor I should note have I linked this to the previous post on the effect of editors and scribes, or changing trends in wordlength of transcripts.

I think there's a lot of things that impact the graphs in this post and I don't mean to suggest war and conflict are the only important factor. It's possible a change in the sources is behind at least some of what we see here. In fact, it probably is.

I wonder if a similar analysis of the London Lives material would support or challenge this post? Or if a look at purely the defendants would avoid the confounding factor of these witnesses or the changing sources.