Published online by Cambridge University Press: 04 February 2015
The late Peter Hart's tentative yet provocative application of the term ‘ethnic cleansing’ to the Irish revolution, especially as practised in Co. Cork, continues to arouse widespread indignation and incredulity. Two associated issues have dominated the debate: the influence of sectarian hatred on the conduct of the I.R.A., and the extent to which Protestants were actually forced out of their homes by intimidation or fear. Though a few historians have endorsed Hart's claim that a substantial number of Protestants were murdered or expelled because of their religion, most argue that such attacks were primarily motivated by political or economic rather than sectarian motives. Similar points have been made, from the revolutionary epoch onwards, by liberal Protestants as well as republican sympathisers, whose rejection of the ‘ethnic cleansing’ hypothesis has sometimes degenerated into vicious ad hominem attacks on Hart and his allies. The issue of motivation is notoriously resistant to historical analysis: all motives are mixed, some are deliberately concealed, and others are unconscious. Little would be gained by making yet another attempt to disentangle the often contradictory hopes, fears, resentments, and rationalisations underlying republican conduct in the revolutionary years.
1 The interminable debate about ‘ethnic cleansing’ among historians and bloggers was prompted by a passage in Hart, Peter, The 1R.A. at war, 1916–1923 (Oxford, 2003), p. 237,Google Scholar somewhat at odds with denials of ‘ethnic cleansing’ at pp 22–3 , 245–6, and 251. This passage first appeared in ‘The Protestant experience in Southern Ireland’ in English, Richard and Walker, Graham (eds), Unionism in modern Ireland: new perspectives on politics and culture (Dublin, 1996), pp 81–98 at p. 92.CrossRefGoogle Scholar Hart’s analysis was anticipated by another Canadian scholar who cited Cork atrocities in arguing that ‘intimidation and violence … played a part in prompting the exodus of the minority’, going ‘a long way to explain the unprecedented flight of the minority during these transitional years’: Bowen, Kurt, Protestants in a Catholic state: Ireland’s privileged minority (Kingston and Montreal, 1983), pp 22, 25.Google Scholar
2 A recent example is Murphy’s, Gerard The year of disappearances: political killings in Cork, 1921–1922 (Dublin, 2010):Google Scholar see my review in Dublin Review of Books (Mar. 2011).
3 See, for instance, Donnelly, James S. Jr, ‘Big house burnings in County Cork during the Irish revolution, 1920–21’ in Eire-Ireland, 47, nos 3 & 4 (Fall/Winter, 2012), pp 141–97.CrossRefGoogle Scholar
4 See Fitzpatrick, David, ‘Ethnic cleansing, ethical smearing and Irish historians’ in History, 93, no. 329 (Jan. 2013), pp 135–14.CrossRefGoogle Scholar
5 For a pithy and cool-headed discussion, placing revolutionary depopulation in the broader context of long-term Protestant decline, see Delaney, Enda, Demography, state and society: Irish migration to Britain, 1921–1971 (Liverpool, 2000), pp 69–83.Google Scholar Recent contributions include Keane, Barry, ‘Ethnic cleansing? Protestant decline in West Cork between 1911 and 1926’ in History Ireland, 12, no. 2 (Mar./Apr., 2012), pp 35–41;Google Scholar Bielenberg, Andy, ‘Exodus: the emigration of Southern Irish Protestants during the Irish War of Independence and the Civil War’ in Past and Present, no. 218 (Feb. 2013), pp 199–233.CrossRefGoogle Scholar
6 The regions of the two states formed through partition are denoted in this article by ‘the South’ and ‘the North’. The names of counties and provinces follow pre-partition usage.
7 For a sympathetic yet balanced account of the demographic and social consequences of the revolution, see McDowell, R. B., Crisis and decline: the fate of the Southern Unionists (Dublin, 1997), pp 163–6.Google Scholar McDowell confirms Kennedy’s view that the precipitate decline of the Protestant population between 1911 and 1926 was unrelated to ‘direct discriminatory actions against Protestants on the part of the new government’ and that Protestant emigration ‘was a voluntary movement’: Kennedy, Robert E. Jr, The Irish: emigration, marriage, and fertility (Berkeley and Los Angeles, 1973), pp 128–9.CrossRefGoogle Scholar This is qualified by Kennedy’s belief that ‘fear of physical harm … accounts for at least part of the one-third decline in the number of Protestants between 1911 and 1926’ (p. 136).
8 The murder of a hundred or so Irish Protestants, often labelled as informers, scarcely altered the aggregate mortality of southern Protestants. Fatalities resulting from service in the Great War would have affected Protestants only slightly more than Catholics, since the majority of wartime servicemen were Catholics.
9 McDermott, R. P. and Webb, D. A., Irish Protestantism to-day and to-morrow: a demographic study (Dublin and Belfast, undated), p. 2.Google Scholar See also Kennedy, , The Irish, pp 110–38.Google Scholar
10 In 1926, when Episcopalians, Presbyterians, and Methodists comprised 7.0% of all occupied males in the South, they accounted for 8.3% of those engaged in defence (including 8.6% of army officers, 7.0% of other army ranks, and no less than 63.4% in 'other defence' categories). But the number so involved had plummeted from 19,621 in 1911 (including retired officers) to 1,264 in 1926 (excluding retired officers). The Protestant sub-group in defence in 1926 would have overlapped substantially with those born outside Ireland, of whom 1,461 had been born in Britain, 38 in the United States, and 95 elsewhere (mainly in the Empire). See Census of Saorstát Èireann (1926), iii, pt. i, Table 16; iii, pt. ii, Table 15; Census of Ireland (1911), county reports.
11 For each major administrative category in the Census of Ireland (1911), the percentage of Catholics was as follows: civil service messengers, etc. (82), police (78), municipal and district council officers, etc. (69), prison officers (68), and civil service officers and clerks (59).
12 The percentage of Catholics in other categories of defence was as follows: Royal Marines, officers and men (31), navy officers, effective and retired (25), and soldiers and N.C.O.s (25). Catholics predominated among pensioners who had resettled in Ireland: army pensioners (69), navy pensioners (70).
13 Among men occupied in defence (except pensioners), the percentage in each denomination was as follows: Catholic (24.4), Episcopalian (63.8), Presbyterian (6.9), Methodist (3.6). The percentage of the entire occupied male population in each of these categories was 74.9, 12.7, 9.8, and 1.3, respectively.
14 My tabulations exclude the small group (1.3%, incorporating those of unstated or no religion) professing religions ‘other’ than Catholicism and the three major Protestant denominations.
15 The problem of accounting for foreign-born Protestants is discussed in Census of Saorstát Éireann (1926), x, General report, pp 46–8; Kennedy, , The Irish, pp 119–21;Google Scholar and Bowen, , Protestants in a Catholic state, pp 21–2.Google Scholar
16 Barry Keane, ‘Ethnic cleansing?’, abridged from ‘The decline of the Protestant population in County Cork between 1911 and 1926’ (unpublished paper, 2012).
17 The census classification for ‘defence’ allows us to distinguish army and naval pensioners, mainly of Irish origin, but not retired officers who likewise belonged to the settled community.
18 Census of Ireland (1911), General report, Table 71. Servicemen were not necessarily co-resident with those returned as their unoccupied dependants.
19 r = +.65 (26 southern counties, accounting for two-fifths of the variance), correlating the non-Catholic component (1911) with the retention ratio for Episcopalians (1911–26).
20 The apparently aberrant case of Kildare, which experienced heavy loss despite its substantial Protestant component in 1911 (13%), is probably a by-product of the unavoidable inclusion of soldiers’ dependants in its base population. Since 44% of Kildare’s Protestants in 1911 were in the armed services, it is likely (on the basis of the returns for Dublin City already cited) that a further 15-20% were their wives or dependants.
1 Hart, , The I.R.A. at war, pp 237–40.Google Scholar In the counties so identified by Hart, the percentage ratio of the Episcopalian population in 1926 to that in 1911 was as follows: Queen’s (79%), King’s and Leitrim (70%), Cork (67%), Limerick (65%), Westmeath (63%), Louth and Mayo (62%), Tipperary (61%), and Galway (48%). Cork’s ratio ranked fourteenth among the 26 southern counties.
22 For trends and fluctuation in cross-border settlement since 1881, see Cormac Ó Gráda and Brendan Walsh, ‘Did (and does) the border matter?’, draft discussion paper for Mapping Frontiers, Plotting Pathways Study Group, no. 16 (2005), pp 5–13, modifying Geary, R. C. and Hughes, J. G., ‘Migration between Northern Ireland and the Republic of Ireland’, appendix to Walsh, Brendan M., Religion and demographic behaviour in Ireland (Dublin, 1970: Economic and Social Research Institute, paper no. 55), pp 37–50.Google Scholar
23 Of male cross-borderers, 44% were currently married, the highest proportions being in Antrim and Londonderry. The corresponding figure for women was 39%, with the highest proportions in Fermanagh and Tyrone. The ‘lost counties’ supplied more married settlers than the other provinces, accounting for 50% of males and 41% of females.
24 Natives of Northern Ireland accounted for only 1.4% of occupied males in 1926, but comprised 11.8% of bank clerks, 8.4% of clergymen, 6.3% of railway clerks, 6.1% of civil service officials and clerks, 4.9% of teachers, 3.4% of textile workers, and 2.8% of railway workers. The high proportions in the army (8.6% of officers, 7.9% of men) and the Civic Guard (4.8%) presumably reflect concentrations of Catholic rather than Protestant Ulstermen: Census of Saorstát Éireann (1926), vol. iii, pt. ii, Tables 14, 15.
25 Census of Northern Ireland (1961), county reports and General report, Table 16. The total southern-born population (53,125, of whom 61% were female) fell about 10,000 short of the figure for 1926.
26 For each county, the sex ratios in 1911 and 1926 were as follows: Londonderry, 158, 157; Antrim, 143, 160; Down, 142, 148; Armagh, 140, 140; Belfast, 134, 137; Tyrone, 127, 139; Fermanagh, 118, 117.
27 Fertility is normally estimated for the period immediately preceding a census from a child-woman ratio involving only infants and women of child-bearing age. Since my purpose is to find a proxy for fertility broad enough to incorporate the demographic experience of a prolonged period of social turmoil, it is necessary to include far broader age-bands than is customary.
28 Gráda, Cormac Ó, Ireland: a new economic history, 1780–1939 (Oxford, 1994), pp 221–2.Google Scholar
29 Note that even if the children of mixed marriages were divided equally between Protestantism and Catholicism, the statistical impact would be more negative for the Protestant child-woman ratio, since the number of mixed marriages would constitute a far smaller proportion of all marriages involving Catholics than the proportion for Protestants.
30 For analysis of religious differentials in fertility in the South from 1926 onwards, see Kennedy, , The Irish, pp 117–18;Google Scholar Bowen, , Protestants in a Catholic state, pp 29–31;Google Scholar Walsh, , Religion and demographic behaviour.Google Scholar
31 The civil registers of births and deaths do not record religion. It is conceivable that centralisation and digitalisation of church registers will eventually spawn comparative studies of Protestant and Catholic family formation, fertility, and the religious division of children of mixed marriages. The first census to cross-tabulate age, marital status, and religion was that for Saorstát Éireann in 1937, which revealed that Episcopalian women were more likely to be married, yet far less fertile, than their Catholic counterparts: McDermott and Webb, Irish Protestantism, pp 17, 27.
32 The ratios were as follows: Co. Dublin, 60; Dublin City, 79; Cork City, 59; Co. Cork, 85. The ratio of adult females to 100 males was 117 in Cork City and no less than 155 in Co. Dublin.
33 r = +.51 (26 southern counties, accounting for one-quarter of the variance), correlating the retention ratio for Episcopalians (1911-16) with the ratio of Episcopalian to Catholic child-woman ratios in 1926.
34 One would expect about 10% of the child population under 10 years in 1926 to have been in their tenth year. If so, the proportions under 9 years in the Free State in 1926 would have been 17.7% for Catholics, 12.1% for Episcopalians, 14.0% for Presbyterians, and 11.4% for Methodists. This would imply deficits (compared with 1911) of 2.8% for Episcopalians, 2.2% for Presbyterians, and 5.1% for Methodists, compared with only 0.3% for Catholics.
35 Non-Catholic marriages registered in Southern Ireland fell from 101 in 1920 to 87, 76, 66, 61, 57, and 60 in succeeding years (1911 = 100). Corresponding index figures for Catholic marriages were 110, 96, 97, 102, 97, 90, and 88: Bielenberg, ‘Exodus’, p. 219.
36 The percentages of male occupied Methodists, Jews, Episcopalians, Presbyterians, and Catholics in each broad category in 1926 were as follows: agriculture, 34, 0.3, 45, 58, 58; producers, 14, 27, 15, 13, 17; commerce, finance, and insurance, 23, 53, 11, 12, 5; public services, 3.7, 0.5, 4.4, 1.8, 3.4; professional, 7.5, 11.5, 6.6, 5.2, 2.4; personal service, 0.6, 0.8, 1.9, 0.8, 1.9; clerks, 9.2, 2.8, 7.2, 4.4, 1.4. See Census of Saorstát Éireann (1926), iii, pt. i, Table 16. Religious differentials in the occupational spectrum (ignoring minor Protestant denominations) are analysed in Kennedy, , The Irish, pp 125–7, 131—4;Google Scholar Bowen, , Protestants in a Catholic state, pp 78–103.Google Scholar
37 Within agriculture (excluding the few Jews), the percentages in each sub-category were as follows: farmers, 57, 49, 50, 39; assisting relatives, 37, 33, 34, 35; paid workers, 6.1, 17.7, 15.9, 25.5. Among farmers, the percentages occupying less than 1-30 acres were as follows : 25, 33, 37, 57. See Census of Saorstát Éireann (1926), iii, pt. i, Table 17.
38 Methodist Church in Ireland, Minutes of Conference [hereafter M.C.M.] (Dublin, annual).
39 r = +.90 (26 counties, excluding Belfast, Waterford, and 5 for which no circuits existed in 1926), accounting for four-fifths of the variance; or r = +.85 (including Belfast and Waterford), accounting for nearly three-quarters of the variance.
40 Annual statistics have been computed for all congregations centred in each region, using figures published annually in the Minutes of the proceedings of the General Assembly of the Presbyterian Church in Ireland.
41 In a passage citing M.C.M. (1911–26), Hart claimed that ‘Methodist membership [in Cork district] was higher in 1918, 1919, and 1920 than in 1914, but fell precipitately thereafter. Once again, 1921–3 were the crucial years, accounting for seventy-four per cent of the lost population’: Hart, , The I.R.A. at war, p. 226.Google Scholar Though it is true that 74% of the loss between 1920 and 1926 occurred in 1921–3, this period accounted for only 30% of the total decline from 1,825 in March 1911 to 1,146 in March 1926.
42 The closure of military and naval stations deprived Irish Methodism of a couple of thousand ‘declared Wesleyans’, but few full members. The last return, for Jan. 1916, enumerated 1,900 Wesleyans in the army (including 587 at the ‘Curragh, etc.’, 388 in Dublin, 122 in Queenstown-Berehaven, 117 in Limerick, and 116 in Cork-Ballincollig), along with 350 in the navy (almost all in Queenstown-Berehaven). None of these were returned as members of the Methodist Church in Ireland. The return for Jan. 1911 gave 962 Wesleyans in the army and 923 in the navy, including 77 and 28 church members respectively: M.C.M. (1916), p. 149; (1911), p. 163.
43 Three circuits listed separately in 1920 were absorbed by neighbours before 1923 (see note to Table 11). The only other southern circuits discontinued between 1911 and 1926, all before 1915, were Mullingar, Killarney (united with Tralee), the Clare mission (no longer returned under Limerick), and Mohill (united with Ballinamore).
44 Manual of the laws and discipline of the Methodist Church in Ireland (Belfast, 1934 edn); Grubb, Arthur Page, A popular handbook of Methodist law and usage (London, 1913).Google Scholar
45 No returns were of ‘removals’ or other components of membership change, with the exception of emigrations and deaths up to Mar. 1919, were published for individuals circuits. The number ‘on trial’ was also returned until Mar. 1920, and the number of junior members in each circuit was published throughout the period. In the North, four circuits returned in 1911 were discontinued or absorbed over the next triennium, and six new circuits were opened in Belfast between 1911 and 1926.
46 Note that the true ‘flow’ was even faster, since the statistics of ‘removals’ measure net movement between North and South, concealing thousands of removals within circuits and between circuits within each region, including multiple removals by individual members. Because the published statistics of inward and outward removals do not specify destination, it is not possible to compute gross movement between North and South.
47 This multiplier assumes that the numerical components of population flow for all Methodist adults approximated those for full members of the church, and that the same rates of flow applied to the entire non-Catholic adult population. Methodists comprised 5.1% of the entire non-Catholic female population in 1911 and 4.9% in 1926 (the same ratio applied to women aged 20 or more). The ratio of church members to the Methodist census population was 74% in 1911 and 73% in 1926, while the proportion of females aged 20 or more in 1926 was 71% for Methodists (as for all non-Catholics). Note that these estimates cannot be extended to children, since the rate of intake of new adult members would only faintly have reflected much earlier patterns of fertility, while mortality, cessation, and emigration rates would normally have been much lower for children than adults.
48 The personal experience of West Cork Methodists during the revolutionary period, and the extent to which they left the region in response to threats or fears, is explored in The spectre of ethnic cleansing in revolutionary Ireland (Cambridge, forthcoming), based on my Parnell lecture delivered at Magdalene College, Cambridge, on 11 Feb. 2013.
49 Returns by ‘religious profession’ are abstracted from the published reports of the Census of Ireland (1911), Saorstát Éireann (1926), and Northern Ireland (1926). ‘Episcopalians’ include adherents of the ‘Irish Church’ or Church of Ireland, the Church of England, the Episcopal Church of Scotland, and ‘Primitive Church Methodists’; ‘Presbyterians’ exclude groups such as ‘Reformed’ and ‘Non-Subscribing’ Presbyterians and Unitarians; ‘Methodists’ include ‘Wesleyan Methodists’, ‘Primitive Wesleyan Methodists’, and ‘Methodist New Connexion’. Figures for ‘Total Population’ include those returned for other religious professions or none. The deduction of those in the ‘Armed Forces’ in 1911 is based on returns of occupied males in each county report, tabulated by religious profession, minus those engaged in ‘defence’(with the exception of pensioners but unavoidably incorporating retired officers; police and prison officers were returned under other categories). The final column gives the percentage ratio of each figure for 1926 to that for 1911.
50 Counties are arranged in descending order by the percentage of females who were not returned as Roman Catholics [F.N.R.C.] in the Census of Ireland (1911). The remaining columns show the % ratio of the population of Episcopalians, Presbyterians, and Methodists in 1926 to the corresponding populations in 1911, excluding male members of the armed forces in 1911 (see note to Table 1). Since county tabulations for ‘Episcopalians’ in Northern Ireland (1926) exclude those returned as ‘Church of England’ (18,682) and ‘Episcopal Church of Scotland’ (41), the column headed ‘Episcopalians’ understates the true % ratios for the 7 northern counties, and the cumulative figures for the North and Ireland. The % ratios for adherents of the Church of Ireland or ‘Irish Church’ alone, using slightly inaccurate county returns in the Preliminary report (1911), are 109 (Antrim, Belfast, and Down combined), 104 (Londonderry), 95 (Tyrone), 92 (Fermanagh), 90 (Armagh), and 104 (Northern Ireland).
51 The cross-border population (C.B.P.) of each county in Northern Ireland refers to those born in Southern Ireland, and vice versa, according to the published Census reports for Ireland (1911), and for the Irish Free State and Northern Ireland (1926). The columns show C.B.P. in 1926; C.B.P. as a percentage of county population in 1926; the numerical change in C.B.P. (1911-26); and the percentage ratio of C.B.P. (1926) to C.B.P. (1911). The counties are arranged in descending order of that percentage ratio.
52 The percentages of Roman Catholics and Episcopalians (excluding adherents of the Church of England) were derived from a digital search of the census online. The sex ratio gives the percentage ratio of Church of Ireland females to Church of Ireland males. The figure for ‘border counties’ aggregates those for Donegal, Leitrim, Cavan, Monaghan, and Louth. The total calculated from the published census reports (56,175) exceeds that derived from digital searches (55,147) by 1.8%, presumably as a result of faulty transcription or standardisation in the digital version.
53 The table shows the ratio of all children under 20 years to 100 women aged 20 years or more.
54 For each province and region, and for each major religious denomination, the table shows the percentage of females aged less than 9 years in 1911 (the only available age-breakdown by religion). The nearest equivalent figures for 1926 relate to those under 10 years (not available for Northern Ireland).
55 Counties are arranged in descending order of the percentage ratio of Episcopalian children under 20 to every 100 women aged 20 or more. The other columns show the equivalent ratio for Roman Catholics, and the percentage ratio of the Episcopalian ratio to the Roman Catholic ratio.
56 Counties are arranged in descending order by the percentage of females who were not returned as Roman Catholics [F.N.R.C.] in the Census of Ireland (1911). The remaining columns show the percentage ratio of the census population of Methodists in 1926 to that in 1911 (excluding members of the armed forces); and the corresponding ratio for the number of full members of the Society returned for circuits centred in each county (M.C.M., 1911, 1926). N indicates counties without Methodist circuits in either year; C indicates counties in which circuits existed in Mar. 1911 but not Mar. 1926.
57 Census returns exclude males occupied in the armed forces (1911). Returns of membership of the Methodist Church in Ireland are abstracted from M.C.M. (1911, 1926), distinguishing between full members received into the Society (excluding Juniors and those on trial) and an aggregate estimate of ‘members and adherents’. This category (used in 1926) evidently corresponds to the figures (returned in 1911) for ‘total not average attendance at public worship on Sabbath’, excluding ‘additional week night hearers’.
58 The districts of Belfast, Clogher, Cork, Dublin, Enniskillen, Limerick, Londonderry, Portadown, Sligo, and Waterford are indicated by the first two letters in each name.
59 The table shows the number of full members of the Methodist Church in Ireland, and the percentage ratio of membership at the end of each period to that at the beginning of the period. Figures for the South and North refer to circuits (not districts) centred south or north of the border established in 1921–2.
60 Aggregate figures for the South and North refer to districts centred south or north of the border established in 1921–2 (returns of flow were not published for individual circuits).
61 Figures are based on annual returns for each church district for the years ending in March, 1911–26 (M.C.M.). Italicised totals indicate minor discrepancies in the published returns. ‘New members’ presumably incorporated reinstatements and promotions from junior membership, but not probationers ‘on trial’ or attendants at services who had not been received as full members of a circuit. Figures for net internal ‘removals’ show the difference between the number of members in each district transferring out of one circuit into another circuit within Ireland, and the reverse (+ signifies net inward transfers; -signifies net outward transfers). ‘Emigrants’ were those ‘removing to any place out of Ireland’, whether or not they joined Methodist congregations in other jurisdictions. The returns also enumerate deaths and the number ceasing to be members for other reasons (such as expulsion, non-payment of dues, resignation, or conversion), completing the elements generating the recorded changes in membership from year to year. Thus ‘net change’ represents the sum of new members and net removals, less the sum of emigrants, deaths, and cessations of membership. Annual rates of flow per thousand members are based on membership in March, 1913 (1911–14), 1916 (1914–17), 1919 (1917–20 and 1911–26), 1922 (1920–23), and 1925 (1923–26).
62 For definitions and sources, see notes to Table 13. Rates of flow are based on membership in Mar. 1919 (1911–26), Mar. 1913 (1911–14), and Mar. 1922 (1920–23).