“The judgement of God on an indolent and unself-reliant people”?: The impact of the Great Irish Famine on Ireland’s religious demography

It has suited both sides of Ireland’s religious and political divide to portray the Great Famine that affected Ireland in the late 1840s as primarily affecting the Catholic population. However, while the geographies of the Famine have been explored in detail in recent years its religious dimensions have been largely ignored, albeit with a few exceptions. As a consequence, the assumption that Great Famine was a Catholic famine has not been sufficiently challenged. Drawing on a relatively untapped source, the 1834 Commission on ‘the state of religion and other instruction in Ireland,’ as well as census data, this paper explore the relationship between religion, poverty and population loss over the Famine period. It shows that Catholics were disproportionately affected by the Famine because the Famine was most severe in areas whose population was overwhelmingly Catholic. In more mixed areas, however, the Protestant population seems to have been at least as affected as Catholics if not more so. This conundrum explains why the Famine actually had a surprisingly small impact on Ireland’s religious demography.


4
The famine that started in Ireland in 1845 and lasted until the early 1850s was arguably both the last major famine to affect Western Europe, and has been claimed to be one of the most devastating famines ever in terms of proportional population loss. 1 The repeated failure of the potato crop, caused by potato blight (Phytophthora Infestans), wreaked havoc on a society that was heavily dependent on mono-crop subsistence agriculture where the crop in question was the potato. The exact numbers who died will never be accurately known, but 1.1 million deaths from a population recorded at 8.2 million by the census of 1841 is a widely accepted estimate. 2 Even this understates the devastation wrought. The twenty-year period between 1821, the first Irish census which recorded a population of 6.8 million, and 1841 had seen Ireland's population grow by 20.2%. Over the subsequent twenty years the population fell by 29.7% as the effects of emigration and reduced fertility compounded the deaths caused by the Famine. The Famine was followed by a long period of continuing population decline and stagnation which lasted well into the twentieth century. Even today, a century and a half later and after the growth and immigration fuelled by the Celtic Tiger economy, the combined population of the Republic and Northern Ireland is still only 6.5 million, below even the 1821 figure let alone the 1841 peak. 3 This makes Ireland almost unique in Western Europe where most countries' populations have doubled or tripled over this period.
Recent years have seen a significant rise in interest in studying the geographies of the Famine. Many of these studies have used census and similar statistics which, more recently, have been analysed using techniques from historical geographical information systems. 4 As described below, these studies have demonstrated that the Famine was more severe in the west, midlands and south. It was particularly destructive in areas that were highly impoverished and that had high growth rates in the immediate pre-Famine period.
There were also religio-political dynamics to the Famine. The mid and late nineteenth century saw rapid economic growth in the north-east of Ireland based around Belfast and the Lagan Valley with their textile mills and shipyards. At the same time, much of the rest of the island stagnated economically and demographically. Given that Ireland's Protestants lived primarily in the north-east and much of the rest of the island was predominantly Catholic, this created a religious, political and economic divide with Protestantswhose manufacturing-based economy was thriving and linked into the wider British and British Empire economiesbeing strongly pro-Union. Catholics, by contrast, increasingly saw themselves as a marginalised, rural population suffering at the hands of British misrule and thus turned increasingly towards Home Rule and Irish nationalism. 5 It suited both sides of this divide to 5 represent the Famine as a largely Catholic phenomenon. To Catholics it was evidence of an uncaring and negligent British state that allowed absentee landlords to make the countryside catastrophically vulnerable to a famine caused by the failure of a single crop, and then failed to deliver effective disaster relief to the stricken population. To Protestants it showed the superiority of their society and its work ethic, and to some it even illustrated their favoured position in the eye of God. The Famine was thus'The judgement of God on an indolent and unself-reliant people.' 6 This popular conception of the Famine as a Catholic experience might be seen to be the residual effect of many decades of nationalist historiography of the crisis, a tradition established by the early genocidal interpretations of John Mitchel. 7 Unionist elites have also been complicit in this process as Protestant suffering during the disaster was antithetical to notions of Protestant deliverance and providence. It might be thought that such notions have been revised out of the historical narrative. However, during the Northern Ireland peace process, Prime Minister Tony Blair expressed regret for the role of the British government in the crisis. This represented a savvy appropriation of the perceived sectarian imbalance of the Famine for contemporary political agendas precisely because the Famine was effectively 'owned' by the nationalist population as 'their' history. 8 Only very recently have historians started to question this narrative or indeed to explore the impact of the Famine on Ireland's religious demography.
[ Figure 1: Ireland's religious geographies in 1834 and 1861] As described in more detail below, sources do exist that allow changes in religious demography over the Famine period to be explored. Gregory et al. present some basic analyses comparing religion in 1834 with 1861.Their basic finding is summarised in figure 1 which compares the proportion of the population of each diocese who were Catholic in 1834 with the proportion for 1861. 9 Intriguingly, it suggests there was very little change. If the assumption that the Famine was a Catholic famine is correct then the Famine should have changed Ireland's religious composition to make Ireland a more Protestant place due to the higher loss of Catholics. This, in turn, should have changed Ireland's religious geographies.
In fact, as Figure 1 shows, the Famine barely seems to have affected these geographies at all.
The trend line shows that the Catholic proportion of the population of each diocese was, on average, almost completely unaffected by the Famine. There are some outliers on both sides of this line: Cashel fell from being 97% Catholic to 90% and Down fell from 31% to 26%. 6 On the other side, Raphoe rose from being 70% Catholic to 77% and Achonry from 95% to 99%. Nevertheless, these changes are relatively minor. The key point is that the change that might be expectedthat the Catholic proportion of the population would decline across the board as the Catholic population was depleted by death and emigrationdid not happen.
Instead, the overall religious geography seems to have been remarkably unaffected, something that can only have happened if the Protestant population was also seriously affected by the Famine. This leads to the central questions that this paper will explore: to what extent was the Great Irish Famine a Catholic famine, to what extent were Protestants also affected, and how did this affect Ireland's religious geographies? BACKGROUND: DEMOGRAPHIC CHANGE DURING THE FAMINE As indicated above, there has been a growing interest in using quantitative and geographical approaches to study the Irish Famine. One of the earliest attempts was Joel Mokyr's analysis of mortality where he found evidence of correlations between excess mortality over the Famine period and poverty and illiteracy prior to it. 10 Perhaps reflecting the technology of the time, Mokyr's was a highly statistical study but was performed at county level and its geographical exploration was relatively limited. More recently, geographical information systems (GIS) have been used to explore the Famine geographically as well as statistically.
Kennedy et al produced an atlas that described the Famine's geographies using data at barony and poor law union level as well as for counties. 11 As there were around 330 baronies and 160 unions they provide a much enhanced impression of the geography of the Famine when compared to counties. O'Grada and O'Rourke and two papers by Gregory and Ell have statistically analysed the relationship between population loss over the Famine period and a range of other variables at these spatial levels. 12 These variables are mainly census variables linked to poverty, particularly fourth class housingthe lowest class of housing counted in the Irish censusesand illiteracy in English. More recently, even more spatially detailed data have become available. Smyth presents an analysis based largely around mapping at the level of the 2,500 parishes, and Fotheringham et al. present an analysis based on around 3,500 electoral divisions and a wider range of variables. 13 While the variables, administrative units and approaches used by these studies vary, taken together they build up a consistent picture of the Famine being worst in the south and west, and of the severity of the Famine being linked to poverty, pre-Famine population growth and dependency on the potato crop. For the small proportion of the entire population living on 7 offshore islands or in close proximity to the coast it may have been the case that access to alternative, littoral food sources may have acted to offset the Famine's worst effects, but even in these zones, such fortuitous dynamics were the exception rather than the rule. 14 Generally, towns and cities had lower levels of population loss than their rural counterparts, and they sometimes even gained population due to internal migration from Famine-stricken parts of the countryside, offsetting the losses that would have been found even in urban areas.
However, such dynamics were negligible and post-Famine Ireland remained a primarily rural society due to the lack of any major industrial concerns outside of north-east Ulster which could attract a major influx of population. 15 While this basic narrative appears increasingly uncontroversial, the issue of the impact of the Famine on religious geographies has been largely overlooked. A consequence of this has been that 'some scholars may unconsciously have repeated contemporary and subsequent claims by Ulster Unionists, who argue that "Ulster"that is, its Protestant inhabitantseluded the Famine.' 16 To challenge this assumption of a lack of an impact on the Protestant population, Miller et al explore the impact of the Famine on ten parishes in east and mid Antrim with high Protestant populations. They show that these parishes had high rates of population loss during the Famine, and are thus able to begin to challenge the assumption of a solely Catholic famine. 17 The difficulty is that their study is only based on a very small number of areas in one of the most Protestant parts of Ulster, and it is therefore difficult to know how applicable its findings are across Ireland as a whole. We thus have a situation in Representatives of all the main religious denominations were also involved and a public meeting was held in each parish to present local results. 18 This painstaking processes led to results that have been described as 'remarkably accurate for an early nineteenth-century statistical study.' 19 Our analysis is limited by the fact that this source is not available in digital form at parish level and digitising it would be prohibitively expensive. The digital data are only available for the Church of Ireland's dioceses. There were only thirty-two of these dioceses and they thus provide relatively little spatial detail. Dioceses further suffer from the fact that, while there are also thirty-two Irish counties, the two geographies are very different and cannot easily be compared even at this aggregate level. Currently, however, this source does provide the best evidence we have of pre-Famine religious geographies and it is incumbent on us to use it to better understand the impact of the Famine. Given that no Ireland-wide analysis of the impact of the Famine on religion has been undertaken, this source provides the potential to allow new insights into how the Famine affected the different denominations and changed Ireland's religious geographies.
Comparing religion in 1834 and 1861 requires dioceses to be compared with baronies. As there were 334 baronies and only 32 dioceses, and that baronies nest fairly well within each diocese, barony-level populations have simply been aggregated to the diocese in which their centroid lay. 20 Exploring the data this produces revealed satisfactory results in all but one case: the diocese of Achonry. Achonry was one of the smallest dioceses with a population in 1834 of only 114,000. Its small area consists of only four complete 1861 baronies plus Costello, 57% of whose area lay within Achonry and the remainder within Tuam. Given that Costello's relatively large population of 46,000 in 1861 was entirely allocated to Achonry this will have over-estimated Achonry's population by around 20,000 people. This represents 17.5% of the diocese's 1834 population. Costello's impact on Tuam is less significant as the diocesan population was 478,000 in 1834, meaning that its population in 1861 is likely to be under-estimated by less than 5%. This seems to have been the only significant error resulting from the aggregation process, caused by the unusual combination of Achnory being small in terms of both population and number of constituent baronies, and one of the constituent 9 baronies with a relatively large population being split approximately in half. An alternative approach to aggregation would have been to use areal interpolation techniques which would probably have given more reliable results in the Achony case but would have made the impact of error more generally more difficult to estimate. 21 The second component of this study is the use of conventional census data to investigate the extent to which the patterns of religious change corresponded to patterns of poverty and population growth. Rather than follow the multivariate approaches used by Fotheringham et al. and Gregory and Ell, this study instead uses an index of disadvantage to draw together several census variables that provide a proxy for poverty. The use of indices of this type, such as Townsend or Carstairs scores, is well established in modern studies of poverty. 22 Carstairs scores, for example, are calculated using four census variables: unemployment, low social class, overcrowded housing and households lacking a car. These are standardised using zscores which measure how many standard deviations from the mean each unit's value is. A unit with a z-score of 0.0 has exactly the same value as the mean, while 1.0 means that the value is one standard deviation above the mean and -2.0 is two standard deviations below the mean. The z-scores for each unit are summed for the four variables, the highest values indicate the highest levels of deprivation. 23 This approach has also been used for historical data in England and Wales 24 but has not, to our knowledge, been used in Irish history.
In this case we compute a similar measure for pre-Famine Ireland using three variables from the 1841 census: the percentage of households living in fourth-class housing, the proportion of the population who could not read or write in English, and persons per building, a measure of overcrowded housing that divides the total population of an administrative unit by the number of occupied houses. 25 As well as the index of disadvantage, we also look at population growth in the pre-Famine period, which we define as between 1821 to 1841, and population change from 1841 to 1861 to look at loss during the Famine and its immediate aftermath. All of these measures can be computed at both diocese and barony level. The barony data provide much more spatial detail than for the dioceses, however barony boundaries changed markedly between 1821 and 1861. To ameliorate the effects of these changes most of the barony-level analyses are done by standardising the data onto a 'target' geography that consists of 302 baronies. These are based on the 1831 baronies with a small number of aggregations to avoid issues associated with boundary changes. They allow us to make comparisons over time without having to consider the impact of boundary changes. 26 One final limitation of our data is that we use the period from 1834 or 1841 until 1861 to represent what we term the "Famine period." This is clearly longer than is desirable. 1845 was the first year in which the potato crop was seriously affected by blight and the Famine is generally seen as ending between 1849 and 1852. This means that our Famine period includes several years of unrecorded pre-Famine population growth, and a post-Famine period lasting around a decade in which emigration and death still caused huge population decline. In the 1840s (1841-1851)the immediate Famine period -Ireland lost 19.8% of its population. The 1850s saw the loss of a further 11.5%, the second highest decade of population loss in Ireland's history whose scale can directly attributed to the immediate aftermath of the Famine. This means that it is not unrealistic to include this latter decade as to be impoverished and that remained sparsely populated even after this growth had taken place. Rapid population growth at this time was not exceptional. Many European countries either were, or would later, go through the same experience. Elsewhere, however, this growth was soaked up by the cities which grew rapidly while rural populations tended to remain roughly static. In Ireland this did not happen. While urban centres such as Dublin, Belfast and Cork, in particular, were growing, these three cities only grew by 154,000 in total between 1821 and 1841, a mere fraction of the 1.4 million by which Ireland's population grew over the same period. The lack of urban areas that could absorb population meant that, in the era that preceded mass emigration, growth remained concentrated in rural areas with the population increasingly reliant on the food that these areas could grow. In Ireland's mild wet climate and poor soils the potato was one of the few crops that could provide sufficient nutrition for the population grow. 27 It should be noted that the apparently low population densities of rapidly growing rural baronies in the south and west may, in fact, be misleading.
In reality, much of the land area of many of these units was uninhabitable. Although it is unmeasurable, the later focus of the Congested Districts Boards (CDBs) suggests that the inhabitable parts of these baronies may actually have had surprisingly high densities. 28 These low densities might thus be argued to be an indicator of just how marginal agriculture was in these areas.
[ Table 1: Correlation coefficients for population growth 1821-41] Table 1  namely that growth was occurring in areas that were sparsely populated and deprived. It is also noticable that the index of disadvantage has a more strongly positive correlation with population growth than its three constituent variables, suggesting that this is doing an effective job of measuring the growing population pressures. [ Table 2: Correlates with pop growth a diocese level] Table 2 investigates the relationship between population growth, deprivation and religion at diocese level in the pre-Famine period. It is important to note that the modifiable areal unit problem means that changing the administrative units used in a statistical analysis will result in changing the correlations found between variables as a result of both the changes in the scale of the unit and changes in their arrangement. In particular, decreasing the number of 12 units used tends to increase correlations. 29 The aggregation of much of the data from 302 baronies to thirty-two dioceses results in both of these issues, but particularly those resulting from the increase in scale. In general, this means that results at barony level should be given more credence than their diocesan equivalents which need to be interpreted with caution. A second issue is that the percentage of Catholics is heavily skewed with twenty-three dioceses being over 85% Catholic and only three being less than 50%. Table 2 shows that the relationship between population growth and population density, fourth class housing and illiteracy remains strong when the data are aggregated to diocese level. The major change is that persons per building is no longer significant at the p<0.05 level, although it remains close with p-values of 0.055 and 0.064 for Pearson's and Spearman's Ranks respectively. A consequence of this is that the index of disadvantage seems less effective at this level of aggregation, although it remains significant at the p<0.05 level. Perhaps more interestingly, using Spearman's Rank, the percentage of Catholics in 1834 is actually more closely correlated with population growth than any other variable, suggesting a very close relationship between population growth and Catholic areas.
[ Table 3: Correlated with Catholics at diocese level] Table 3 shows that the Catholic population is very closely correlated with the index of disadvantage, with the exception of persons per building, but is less closely correlated with population density. In summary, in the immediate pre-Famine period much of Ireland had a population that was overwhelmingly Catholic. These areas, concentrated in the west and south, tended to have high levels of deprivation and high levels of population growth. Not all areas with high Catholic populations had high population growth, but it is striking that all of the dioceses with high population growths were also overwhelmingly Catholic. Drawing from the existing literature, this would suggest that the Catholic population was much more vulnerable to the Famine as the areas that could be predicted to be most vulnerable to it also had very large Catholic majorities. This will be investigated in the next section.

THE IMPACTS OF THE FAMINE
[     Tables 5 and 6 show the barony-level and diocese-level correlation coefficients between population change over the Famine period and population density, pre-Famine growth, the index of disadvantage, and the three variables that make it up. The confirm the link established in the literature between population loss and poverty, with illiteracy in 1841 being a particularly good predictor of subsequent population loss. Variables associated with pre-Famine population change and population density are less significant, which is perhaps a little surprising. Aggregating these data to diocese level, seems to confirm these patterns. The interesting thing in these figures is, however, the strength of the relationship between the proportion of Catholics and population loss which, at 0.745 or 0.711 respectively, is stronger than any other relationship. This suggests that Catholics were more severly affected by the Famine than even their preponderance in vulnerable areas would lead us to predict.
[ Figure 5: Caths 1834 and Cath change] We have thus established that Catholics were more seriously affected by the Famine than Protestants because they lived in areas that were more vulnerable. There is, however, a risk that simple correlations do not convey the complexity of the patterns and relationships that are occurring. Figure 5 shows a scatterplot that compares the Catholic population in 1834 with the subsequent decline in the Catholic population until 1861. There is a correlation between these two variables of -0.420 or -0.429 using Pearson's and Spearman's Ranks respectively which are both significant at the p<0.05 level.
This, however, only tells part of the story. Rather than explore the linear relationship between the two sets of values, it is possible to identify three different clusters of dioceses. First, and most obvious, are those areas where a high proportion of their populations were Catholic in 1834 and which subsequently suffered a high decline in this population. This cluster might be bounded by being at least 80% Catholic in 1834 and having subsequent declines of over 15%.
Twenty-two of the thirty-two dioceses would fall in this cluster, which experienced an average decline of 36% from a population that was, on average, 94% Catholic. The second cluster is those that have lower Catholic populations but that still experienced a high loss of Catholics. These dioceses have fairly mixed religions with Catholic populations between 30 and 66% but still had losses of over 15%, indeed these losses average 28%, which is lower 15 than the more strongly Catholic areas but nevertheless still high. The third cluster is those that had low Catholic losses but whose Catholic populations vary widely. These have losses of less than 6.5% but their Catholic populations vary from 26% to 95%.
[ Figure 6: Cath clusters] These clusters are mapped in Figure 6. This shows that the majority of Ireland had a high This would not explain Raphoe in the north-west, which approximates to County Donegal, and Achonry to its south. Achonry can probably be discounted due to the data issues discussed above. Raphoe is more difficult to explain. It has been suggested that proximity to the coast allowed famine victims to fish and it may be that this made this relatively coastal diocese resilient to the Famine. 32  Figure 9 reveals that this national trend was rarely followed more locally. Indeed, in only two dioceses, Ardagh in the north Midlands and Leighlin south-west of Dublin, do the actual pattern of losses broadly follow this pattern.
For most of the rest of the country there were either significantly more Catholics lost than would be expected, or significantly more Protestants than expected. As with the previous analysis there is a clear geography to this. Away from Ulster and Dublin, Catholic losses were much higher than would be expected given the national pattern. In all but one diocese, Meath, these losses were over twice what would be expected and in four -Tuam, Kilmacduagh, Ardfert & Aghadoe and Cashel -Catholic losses were five times higher than would be expected from the national ratio. In two of these -Kilfenora and Emlythere was actually a gain of Protestants alongside a major loss of Catholics although, as was established above, this is against a backdrop of very small Protestant populations in these two dioceses.
Thus the areas where loss either reflects the national aggregate or where Catholic losses were higher than would be expected (often significantly higher) were found in the south and west in areas that were overwhelmingly Catholic. 4,438,000 Catholics were enumerated in these dioceses in 1834 compared to only 247,000 Protestants or, to put it slightly differently, these areas contained 71.2% of the country's Catholics and only 16.5% of its Protestants. Therefore, areas with very high Catholic populations were not only disproportionally heavily affected by the Famine, their losses typically fell very heavily on their Catholic populations because their Protestants tended to be more affluent and insulated from the Famine's effects.
The remaining dioceses are either in the north or on the east coast. Across the three dioceses of southern Ulster -Kilmore, Clogher and Armaghand in Ferns, south of Dublin, there was a higher loss of Protestants than the national ratio would have us expect. In Derry and Down approximately even numbers of Catholics and Protestants were lost. In 1834 Derry was 53.4% Catholic while Down was 31.0% Catholic, suggesting that in these parts of Ulster the ratio of losses between Catholics and Protestants was close to one to one in Derry and weighted towards more Catholic losses in Down. In Waterford, Dublin, Dromore and Raphoe the number of Protestants lost exceeded that of Catholics. Finally, Connor is again exceptional, experiencing a slight loss of Protestants (2.7%) and a modest gain in Catholics (7.6%). This can probably be largely explained by in-migration into Belfast, particularly by Catholics, combined with a loss of Protestants from the more rural parts of the diocese. This shows, therefore, that where there were significant Protestant populations these were often at least as badly affected by the Famine as Catholics. The nine dioceses that were more than 15% Protestant in 1834 contained 81.6% of the Protestant population. These dioceses experienced a 15.9% decline in their total populations between 1834 and 1861 which, while not quite as high as the 27.7% losses experienced across Ireland as a whole, was still little short of catastrophic. Importantly, these losses were almost evenly divided between the two religions: the Catholic population dropped by 16.9% while the Protestant population fell by 14.5% suggesting that in these areas the Protestant population was as vulnerable to the Famine as Catholics. This was not, however, caused by indolence, but instead by a lack of the industrial growth 19 required to soak up the excess population, a theme that led to a continuing culture of mass emigration from Ireland in the nineteenth and twentieth centuries. The judgementwhether it came from God or from Phytophthora infestanswas not sectarian, instead it reflected the different geographies and socio-economic statuses of the two denominations. Far from being a Catholic famine, the Great Irish Famine was a famine of the rural poor. Over much of Ireland this group was predominantly Catholic, and thus the Catholic population was disproportionately affected. However, the impact on Protestants increased in areas with larger Protestant populations to an extent that in mixed areas it is impossible to say which denomination was more severely affected. As a result, the Famine and its immediate aftermath did not result in major changes to Ireland's religious geography. This matters. The Famine remains a defining catastrophe in Ireland's history and has an enduring power to reinforce the stereotypes from which both communities continue to construct their own selfidentities. This paper shows that the experiences of the two communities were more similar than either would tend to assert.        22