The secret code of Irish names

Anna Burns’ Milkman is a remarkable book. Based on her experiences growing up in Belfast during the Troubles, it conveys a claustrophobic sense of oppression and suspicion, gossip and control. Words and actions, even objects, are burdened with an oblique symbolism that draws meaning from its sectarian context – the "us" and "them" against which the book's disconcertingly mundane plot plays out.

Given this weight of symbolism, maybe the most remarkable feature of Milkman's prose comes from Burns' decision to strip the most basic symbols of all – proper names – from the text almost entirely. Characters are denoted with phrases such as "middle-sister", the narrator and main character. For groups and places, "the other side of the street" and "over the water", stand in for "them" (the Protestant community) and Great Britain respectively.

I won't dwell on the literary effects of this choice here – Roger Gathmann has provided an excellent analysis of this already. Instead I'll focus on one particular list of proper names that does make it into the book: the "banned names list".

This is a list of 69 first names, all male, starting with Nigel and Jason (also the nicknames given to the couple who maintained the list), which were understood to be too British and therefore not allowed within the narrator's own community:

The names not allowed were not allowed for the reason they were too much of the country ‘over the water’, with it no matter that some of those names hadn’t originated in that country but instead had been appropriated and put to use by the people of that land.

This artefact of imagined folklore relates to a real phenomenon – the uncanny ability of anyone who has grown up in Northern Ireland to discern whether someone is from the Catholic or Protestant community just by their name. As someone who didn't grow up in "this place" (another denoting phrase), I am still amazed to see my wife and her friends (who did) demonstrate this ability with some confidence. Could they really be right every time?

A faithful encoding

I decided to investigate the basis for this claimed skill, using Milkman's banned names list as a starting point and historical editions of the Irish census as my investigative tool. This census is exceptionally rare amongst its global counterparts in exposing detailed records for the entire population at individual level. In particular, names and religious affiliation are recorded for virtually everyone who lived in Ireland at the time of the surveys.

This suggested a clear experimental design: If the relative popularity of some names was starkly different for one community (proxied by religion) compared to the other, this could explain the collective societal knowledge that allocated the appropriate label to each name.

My investigation into the religious encoding of these names led to further onomastical discoveries, including the first inklings of the Gaelic revival in baby naming and the surprisingly low literacy of people with the surname Joyce. Read on for my analysis, and if you like what you read, please consider subscribing below.

The Irish census – an exceptional resource

A systematic government census of the whole island of Ireland was taken every 10 years between 1821 and 1911. Of these, the 1901 and 1911 editions are the only surviving full censuses. Following partition, in 1926 two separate censuses were conducted north and south of the new border. However only the records from the southern 26 counties have survived from this year. Incredibly, it appears that the records for the six northern counties were pulped during WW2 in an effort to alleviate a wartime paper shortage.

Privacy laws in Ireland allow the full individual-level census records to be made available 100 years after the survey. As of April 2026, the 1926 edition, as well as those for 1901 and 1911, is now available. However, because only these earlier censuses contain records for all 32 counties of Ireland, North and South, the 1911 survey forms the basis for most of the analysis here.

Census returns contain details about the population in varying degrees of fidelity and completeness. Names themselves are virtually complete. The other fields I will be making extensive use of are religion, education (which in practice encodes literacy) and county of residence. These are well populated. Other fields such as occupation are not considered here, mainly due to their sparseness.

Does census evidence tally with Nigel and Jason's list?

My first question concerned the names on the banned list. As Burns described in the quote above, it was not the origin of names that led to their ban, but rather their popularity and usage amongst members of the "other" community.

This is something the census can measure precisely, and although Milkman was set in the 1970s, I reckoned that the 1911 census would still provide a useful tool to investigate the relative historical prevalence of these names.

Some of the names on the list are not really names ("Duke Of", "Sir Mary"), and others are vanishingly rare. Ten of the 69 don't appear in the census returns at all, four appear once, and a further 14 (including Nigel, a pointed reference perhaps) appear fewer than ten times.

But 23 of the names do appear at least 100 times – frequently enough to indicate their relative popularity in each community compared to the other. Statistics for these names are plotted below:

This number-line displays two attributes for each of the 23 names that occur 100 times or more: the size of the circles represents the overall popularity of the name, whilst the position along the line shows the percentage of people with that name that identified as Protestant. The circles can be hovered over or clicked to bring up more information.

I have also plotted a vertical red line which displays the overall religious mix of the all-island Irish population at that time. From this we can infer that names to the right of the red line were relatively more popular among Protestants. Names to the left were relatively more popular among Catholics.

For these 23 names, there is a clear division: 15 of the names were not only more popular amongst Protestants, this was true to such an extent that there was an absolute majority of Protestants amongst people called them, even in a country where nearly three-quarters (74%) of the population was Catholic. The most popular of these names was George, accounting for 25,944 Irish people in 1911, 15,487 of whom were Protestant.

In contrast, 8 of the banned names were actually significantly more popular amongst Catholics in 1911, including Simon, Felix and both variations of Laurence. So why are they on the banned list? Maybe they had shifted in their relative popularity by the 1970s, possibly after being popularised by a British celebrity. More likely, Burns was just exercising her creative licence by including them. They certainly didn't strike me as out of place as I was enjoying the book.

One thing we can say is that these names, more popular in the Catholic community, are hard to disentangle on the number-line above, as they are bunched towards the left-hand side. Our analysis could also be made more intuitive if the red line of equal popularity was placed in the middle of the chart.

The chart below addresses these issues. It displays the same information, but this time the horizontal axis shows relative popularity (as inferred by relative prevalence) of each name:

This transformation yields easier interpretation – names to the left of the red line at x=1 (indicating equal popularity) are more popular in the Catholic community, whilst names on the right are more popular amongst Protestants. Again, the circles can be hovered over or clicked to bring up more information.

Extending the analysis to all names

We now have a tool to gauge the relative popularity of names by community, which we can use to examine the full set of names, including female names and surnames. But it's obvious from the charts above that fitting even 23 names on a one-dimensional line leads to significant clutter.

One way to overcome this problem and accommodate more names is to introduce a second dimension, which has the further benefit of allowing us to examine another characteristic. After some brief experiments with other options, I decided to use the responses to the "education" question for this purpose.

This question asked respondents whether they could read, read & write, or neither. It was asked independently of a second question on language ability (with responses English, Irish or both). I decided that the responses to the first question could be used as a good proxy for literacy in either language, and that this measure could give some insight into social class.

Overall adult literacy in Ireland in 1911 was fairly high at 88.8%, but there was significant variation between names. In an economy that was still dominated by agriculture, this variation likely correlated with families' ability and desire to educate their children for more middle-class professions, as well as to participate in Ireland's rich literary culture.

Filtering out respondents below the age of 15, I computed the adult literacy rate (combining the "read" and "read & write" categories) for each name. The chart below shows the 20 most popular names of each sex for each community:

The horizontal axis on the chart above stays the same – it shows the relative popularity of first names by community, with the vertical axis at x=1 (now grey) indicating the threshold of equal popularity. Adult literacy rate is indicated by vertical position in the chart, with the average rate for Ireland (88.8%) indicated by the vertical position of the horizontal axis, dashed in grey.

The axes therefore split the names into four quadrants: those that are more popular amongst Catholics (left) vs Protestants (right); and those that have a higher-than-average adult literacy rate (top) vs a lower-than-average rate (bottom).

The first thing to note is that the size of the circles has changed. George is still present as one of the most popular names, but it is dwarfed by Mary and John which each numbered around 300,000 people in 1911, together representing 15% of the total population.

Perhaps unsurprisingly given the wealth divide at the time, there is also a fairly clear correlation between the variables, with the names more popular amongst Protestants having higher adult literacy rates on average. Indeed Hugh is the only one of these popular names that was more popular amongst Protestants but less literate than average.

There are some features that did surprise me though. I had expected the name William to be much more clearly Protestant, given its association with "King Billy" and the battle of the Boyne. In fact, the name was only twice as popular amongst Protestants, meaning there were actually more Catholics named William in 1911, owing to the greater overall size of the Catholic population.

I was also surprised that Bridget, Michael and Patrick, very common names in 1911, were almost exclusively Catholic, showing a greater degree of polarisation than any popular Protestant name of the time.

As for literacy rates, it is interesting to note the differences between Anne and Annie, and between Catherine, Kate, Katie and Kathleen. These are significant differences between names that we might otherwise suspect were interchangeable – in fact today's prejudices might even lean the other way, with a Catherine as the current Princess of Wales possibly marking this name out as the poshest (most likely to be literate?) variant. Nevertheless, in 1911, Catherines were significantly less literate than Kates, Katies and Kathleens.

First name frequencies as evidence of the Gaelic revival

Aside from the two dimensions plotted in the chart, a striking feature of the names it contains is that all their spellings look British to the modern eye. There are no distinctively Irish names, as you might expect if we were using the latest census data.

In fact, even now-common names such as Aoife and Sean (or Seán spelled with a fada to lengthen the "a") were very rare in 1911. This name – the Irish version of John – only appears 124 times in that year's census (97 with the fada, 27 without). Adding three further Gaelic variants of the name – Seaghan, Seághan and Seaghán – brings the total to 911, but this is still 300 times less frequent than the Anglicised "John", the most common male name in 1911, which appeared 288,954 times.

Those 911 Seáns represent significant growth since the 1901 census however, when the name appeared just 10 times in any guise. And by 1926, in the 26 Southern counties alone, it had multiplied in popularity yet further, appearing 3,113 times in total. This growth rate is clear evidence of the Gaelic revival – a national revival of interest in the Irish language and Irish Gaelic culture that had begun in the late-nineteenth-century.

By using a "regular expression" – a pattern matcher – we can match many more Gaelic names to build a more statistically robust measure of the growth in Gaelic naming over this period. To do this I used a regular expression that matched any name containing a fada, as well as the common Gaelic digraphs "bh", "dh", "mh", "fh" and the trigraph "aoi", which are rarely found in English names.

This revealed strong evidence of the revival. In 1901, a total of 265 names in the whole of Ireland (0.006%) contained a fada or one of these substrings. By 1911, that number had risen to 12,071 (0.3%), and by 1926 it had risen to 24,149 in the Southern 26 counties alone, representing nearly 1% of the population at the time.

This growth, the first shoots of which we have detected here, has only accelerated since – apply this filter to the latest baby name data from Ireland's Central Statistics Office today (for babies born in 2025) and over 15% of names match.

This is a fascinating finding, and the distribution of first names yields many other avenues to explore, some of which I pursued briefly in the research for this article. However, a full exposition of those avenues will have to be reserved for a follow-up article, as we have still to apply the tools developed here to surnames. If you want to explore the first name data in more detail yourself, please see the tables and further resources at the end of the article.

Surname analysis – some surprising discoveries

Turning our attention to the storied family names of Ireland then, we can ask some interesting questions. Which clans were more insular in their own religious community, and which mixed more? Which were more literate, and how were they geographically distributed?

The chart below plots the 150 surnames that were most common across the island of Ireland in 1911, on the same set of axes we used for first names:

The correlation we noted before is still present, albeit less obvious – of the 37 surnames in the top 150 that were more prevalent amongst Protestants in 1911, only 3 (in the lower-right quadrant) had lower literacy than average, compared to 68 of the 113 that were more prevalent amongst Catholics.

Intriguingly, these statistically more-Protestant but less-literate names included Burns, now a Booker prize-winning name thanks to Milkman, and McBride, one of the most recognisable names amongst Northern Irish writers today.

Yet more surprising is the revelation that, of the top 150 surnames in Ireland in 1911, the one with the lowest adult literacy rate of all (and the only one to have a rate below 75%) is none other than Joyce, possibly the most lauded name in Irish literature at the time. Even James Joyce's absence from the census records due to his residence in Paris at the time can't explain that.

There are likely many more stories hidden in the chart above that could be unlocked with the relevant context, and I plan to revisit this data for further articles. For now though, I'll leave it to you to interact with the chart to explore the other 147 names. As with the first name data, I've also included a comprehensive table containing surname data at the end of this article – please explore them and get in touch with your observations and thoughts.

The data behind the tea towel

The final part of my investigation uses another field that is reliably captured by census surveys – the county where people lived. Here I use this data in two ways: firstly, I apply the approach used above to counties, displaying their relative religious mix and adult literacy rates; secondly, I combine this field with surnames to map out the dominant clans in each county, a bit like those souvenir tea-towels you might have seen but with better data. I've chosen to use the 1911 census for this analysis too, as it allows us to render all 32 counties.

Plotting these data against our two now-familiar axes gives us the following plot:

Looking at the horizontal axis, the rationale for splitting Ulster in pursuit of a reliably Protestant-majority Northern Ireland can be discerned. While Fermanagh and Tyrone had narrow Catholic majorities, the new province needed to retain their populations (and land area) to be a viable entity. Retaining Monaghan, Cavan and Donegal would have kept a further 330,000 people, but 78% of them were Catholic – too many to ensure a stable majority. As we now know, Northern Ireland's Protestant majority would not last anyway, despite this gerrymandering.

Away from Ulster, there was significant variation in the size of the Protestant population. Cosmopolitan Dublin and neighbouring Wicklow and Kildare had relatively large Protestant populations, whereas the smallest were found on or near the Atlantic Coast in Mayo, Roscommon, Galway, Kerry and Clare.

As for the vertical axis, Dublin, Antrim and Down had the highest adult literacy rates, likely driven by Dublin city and Belfast, the economic and cultural centres of their day. At the other end of the scale, Donegal sat stranded from most of the other counties, an outlier in both a literal and literate (not to mention religious) sense.

Mapping Ireland's ancestral dynasties

Our last chart combines the census returns for surname and county to map the most common family names for each county:

This map highlights the regional dominance of certain names. Kelly is on top throughout Ireland's central belt from Kildare to Galway, whereas Byrne has a stronghold in Dublin and the counties to its south. A full 7% of the population of Tipperary was named Ryan in 1911.

Other names are more spread out, such as Murphy, the most common name in Cork, Wexford, Louth and Armagh, as well as in Ireland as a whole, with the 55,796 Murphys recorded in the 1911 census representing 1.3% of the entire island's population at that point.

What's in a name? More than first meets the eye

What started as an investigation into the curious ability of Northern Irish people to categorise each other by name alone led me to several further discoveries, and an appreciation that there is often more to a name than first meets the eye. I hope you found some of these discoveries as interesting as I did.

If you would like to explore the data further, you can either access it directly via the github page for this article, or explore the tables below which contain more information on first names, surnames and counties that I couldn't fit into the charts in the main article.

Finally, if you have any questions or suggestions for further analysis, please do get in touch. Thanks for reading!

Reference tables

Searchable tables of first names, surnames and counties from the 1911 census are displayed below. The code to generate the data and analysis behind the charts in the article can also be found on github.

Please explore, download the data using the links at the bottom of the tables, and get in touch if you have any questions.