Surname Diversity in Great Britain
As part of my PhD research I recently produced the map below (high res. version) that shows the diversity of surnames in Great Britain. I wanted to demonstrate that surname diversity is not uniform across Great Britain. For example towns and cities (especially London) have relatively high surname diversities compared with rural areas because more migrants and single people live in them. Wales has a very low surname diversity due to its past naming conventions. The measure used is calculated by dividing the number of surnames by the total population of each Output Area (OA). There are over 200,000 OAs in Britain. Urban OAs are very small despite accounting for a large proportion of the total population, so I have scaled the size of each OA by their population (the map is therefore a cartogram). This creates the somewhat bloated appearance of Great Britain, but serves to emphasise the increased surname diversity (due to more single people and migrants) in towns and cities. The correct shape of Great Britain is shown in the inset. For more technical info please see below.
To create this map I used ArcGIS 10 and the Cartogram Geoprocessing Tool. The nice thing about the tool is that it is not dependent on VBA and therefore worked straight off in ArcGIS 10. There are over 220,000 spatial units in this map and the tool had no problems processing them. I have not found any alternatives that work for this volume of data.
Great Britain’s Surname Cloud
As part of the research group that created the National Trust Surnames Profiler I have access to a comprehensive database of surnames in Great Britain. Most of my analysis on this database is spatial so I thought I would look at non-spatial way of illustrating the fact that in Britain the majority of people have a rare surname and that there are relatively few “popular”. This distribution is often referred to as having a long tail, as can be seen from the graph below. I find this graph lacks impact and novelty and it is hard to label a meaningful number of names along the x-axis. The surname clouds above have a greater impact by clearly demonstrating the popularity of a few surnames, such as Smith and Jones, in Britain and the large numbers of rarer surnames. I have only included the top 500 names for clarity. Comparisons between 1881 and 2001 are interesting. It would appear that the big names, such as Smith, dominate less now than they once did. The effects of migration also show in 2001 with names such as Patel, Kahn and Singh making an appearance. You can see how your name compares globally here. I am not sure if a word cloud would stand up to peer review for a journal but I think it would make a more interesting addition to presentations and posters than a simple line graph.
BBC Radio Wales: Welsh Surnames
I recently had an interview with Radio Wales‘ “Good Evening Wales“. Following media interest in the migrations of some Welsh surnames such as “Jones”, BBC Wales wanted a little more information from our own World Names Profiler project. The project’s website allows visitors to type in their own surname and generate a map of it’s global distribution. You can also do an ethnicity search to simply map where Welsh names, for example, occur (see map below).
From the website you can find some interesting facts. For example, you are more likely to meet someone with a Welsh name in Chicago than London, and 6 out of the “top ten” regions with the most Welsh surnames (outside of Wales) occur beyond Europe. One of the most successful migrations (in terms of preserving the Welsh language and culture) was of course to Patagonia and this is shown by Argentina appearing in the top 10 most likely places to find many Welsh Surnames. So, although the main focus of yesterday’s interview was the movements of Welsh surnames within the UK, I think the global migrations we can track using Welsh surnames are far more interesting.
That said, to illustrate a little more the media interest in the Welsh surnames within the UK, I recommend people visit the National Trust Surname Profiler Website (link) that provides historical and contemporary maps of most surname distributions in the UK. The data behind this website have been the focus for much of my research and I have produced some maps related to Welsh names already. I have and included a couple with a little commentary below. If you would like to make your own you can visit the websites I mentioned above (Worldnames, National Trust).
The map above shows the % of the population with a Welsh surname (left) and an English surname (right). Darker brown means higher percentages and lighter colours represent lower percentages. You can see clearly how the more urban Southern Wales and the Welsh border have been infiltrated with English surnames.
I have featured the above map before on this blog. I have rescaled the UK so that the size of the area is proportional to the number of people with the Welsh surname “Lewis” that live there. As you can see from how much larger Wales has become you are still most likely to find the Lewis name in its country of origin.
IEEE Geospatial Computing Workshop
Yesterday I presented the paper “Combining Historic Interpretations of the Great Britain Popualtion with Contemporary Spatial Analysis: the Case of Surnames” during the Geospatial Computing Workshop at the 5th IEEE International Conference on e-Science . You can download the extended abstract here and I have uploaded the complete presentation below. In later posts I will provide a summary of the other papers presented in what I thought was a very interesting session.
CASA Working Paper
My Centre for Advanced Spatial Analysis (CASA) working paper titled “Family Names as Indicators of Britain’s Changing Regional Geography” has been published online. The paper is one of my PhD upgrade documents and contains the results from much of the research I have completed in my first year. I begin by outlining the significance of surnames in Britain before outlining some of the spatial analysis I have employed to discover if there is a regionality in the British populations’ surnames, and whether this has changed between 1881 and 2001. My results contain plenty of maps to demonstrate where I think the regions fall and I accompany these with a discussion of the appropriateness of each of the methodologies I have used. I must stress that this is a working paper and therefore provides a benchmark on the way to completed research- many of my results and the subsequent conclusions have yet to undergo the rigours of peer review. You can download a copy of the paper here. I have pasted the abstract below.
Family Names as indicators of Britain’s Changing Regional Geography
James Cheshire, Pablo Mateos, and Paul A. Longley
In recent years the geography of surnames has become increasingly researched in genetics, epidemiology, linguistics and geography. Surnames provide a useful data source for the analysis of population structure, migrations, genetic relationships and levels of cultural diffusion and interaction between communities. The Worldnames database (www.publicprofiler.org/worldnames) of 300 million people from 26 countries georeferenced in many cases to the equivalent of UK Postcode level provides a rich source of surname data. This work has focused on the UK component of this dataset, that is the 2001 Enhanced Electoral Role, georeferenced to Output Area level.
Exploratory analysis of the distribution of surnames across the UK shows that clear regions exist, such as Cornwall, Central Wales and Scotland, in agreement with anecdotal evidence. This study is concerned with applying a wide range of methods to the UK dataset to test their sensitivity and consistency to surname regions. Methods used thus far are hierarchical and non-hierarchical clustering, barrier algorithms, such as the Monmonier Algorithm, and Multidimensional Scaling. These, to varying degrees, have highlighted the regionality of UK surnames and provide strong foundations to future work and refinement in the UK context. Establishing a firm methodology has enabled comparisons to be made with data from the Great British 1881 census, developing insights into population movements from within and outside Great Britain.
Cartograms
Last week the BBC News Magazine featured an article on cartograms. The cartograms were produced by the University of Sheffield’s Geography Department and originally published on their Worldmapper website. I find the maps easy to understand and especially effective at representing global inequalities such as water use, but they also work well for population distributions and are most often used in elections to show the relative influence of each election district on the election outcome. Cartograms are produced by placing a regular grid over a conventional map of whatever it is you want to show- such as population- and then stretching or shrinking the grid cells based on the variable (in this case population) of interest. A good analogy would be drawing a map of the UK on a balloon and stretching it to distort the size of southern England in relation to Wales if you want to show its high population density. Simple distortions in relative size have the limitation that the outlines on the regions of interest have to change in order to match up and remain contiguous. This makes the regions unrecognizable and the maps hard to interpret. Gastner and Newman solved this problem (technical article) through some fairly advanced physics, so now we can create cartograms that keep the outline of the regions we are distorting consistent and therefore much simpler to interpret. Tom Gross has created a Cartogram Toolbox for ArcGIS, whilst ScapeToad is very good free cartogram software. Using the former I have taken some of my surname data and produced the cartograms below. They provide a different and, in some cases, much more powerful method of highlighting the spatial distribution of names in the UK. You can compare the effectiveness of these maps with those produced by the National Trust Names website.
This is a cartogram representing the relative frequency of the surname “Cheshire” in the UK. You can see it is an English name with very few, if any, occurrences in Scotland, Wales and Ireland. West Central England has become the most bloated suggesting the highest proportion of “Cheshires” in this region. I have mapped a silhouette of a correctly proportioned (British National Grid) UK behind the cartogram representation.
The above cartogram shows the surname “Lewis”. Lewis is a common name throughout the UK but the cartogram shows its Welsh (particularly South Wales) dominance as the country has greatly expanded beyond its normal proportions.
This is my favorite cartogram. It shows how concentrated many Cornish names, in this case “Laity”, are in the UK and how they rarely exist beyond Bristol.
The above cartogram is for the surname “Saltmarsh”. I have included this as it nicely represents the influence of landscape features on UK naming conventions. The sheltered South Eastern and Eastern coast of the UK is where you will find the majority of UK saltmarshes. The cartogram shows an expansion of districts on the South Eastern/ Eastern side of the UK and a reduction in district size along more exposed coasts (where there are fewer, if any, saltmarshes), therefore suggesting a close correspondence between the number of people with the name and their proximity to the landscape feature!
Featured Map: “Non-British Names Map”
I wanted to avoid the misleading effects of poorly selected spatial units and inappropriate data categorisation in my next Featured Map. I produced it for a forthcoming CASA Working Paper to inform my analysis of general surname trends in Great Britain by accounting for areas of large numbers of “non-British” names. By running Electoral Roll data through UCL’s Onomap classification I was able to classify people’s names according to their origin, and extract all the names that were not classified as “English”, “Scottish”, “Welsh” and “Cornish”. This method is, of course, not completely perfect but it provides very good context to my analysis. I have opted for 10 categories and small spatial units (Output Areas) to emphasise the clustering of “non-British” names around urban centres. Reproducing the map at Admin. District level gives the impression that the most of Britain is inhabited by people with “non-British” names- something that is clearly not the case.
Featured Map: Possible Surname Regions in Britain
This map is one I created for a working paper I am currently writing. Using surnames from the 1881 census of Great Britain, I have been implementing a measure (the Lasker’s Distance) that establishes the similarity of populations based on their Coefficient of Isonymy. The Lasker’s Distance enables the similarities or differences between populations (in this case those in British Districts) to be thought of in terms of ‘surname space’. Put simply, areas that have more surnames in common can be considered closer together than areas with fewer common surnames. This is useful as it enables a number of analysis techniques to be applied to the resulting distance values. In this case Ward’s Hierarchical Clustering has been applied to produce 15 clusters. The map represents each of these clusters with a unique colour. These results suggest that regions are present in the geography of British surnames. This map should be considered as preliminary as it is yet to undergo academic review.
POPfest 2009 Presentation
Below is the presentation I gave at POPfest 2009 on the 2nd July. The conference took place at LSE during one of the hottest weeks of the year and the organizers did a great job of keeping the delegates happy with plenty of strawberries, chilled drinks and sweets. I found the informal and friendly atmosphere a great place to discuss new ideas and find out what other postgraduates are up to in the field of Population Studies.
