THE NORTHERN IRELAND PARLIAMENTARY ELECTORAL SYSTEM : THE 1929 REAPPORTIONMENT

THE NORTHERN IRELAND PARLIAMENTARY ELECTORAL SYSTEM : THE 1929 REAPPORTIONMENT ROBERT D. OSBORNE School of Sociology & Social Policy Ulster Polytechnic ABSTRACT Electoral geography is a growing branch of research in human geography. The study of the spatial organisation of elections constitutes one of the main strands in this field. In this paper the transition in 1929 from a proportional representation (single transferable vote) system to a plurality electoral system for Northern Ireland parliamentary elections is the focus of attention. The motives for the alteration and the method of reapportionment are assessed and the new spatial framework tested for malapportionment and gerrymandering. Finally, the politico-geographic consequences of the reapportionment are examined. Electoral geography has become an area of increasing research activity in human geography. A number of distinct strands in this burgeoning field can be distinguished 1. One such strand has developed around the spatial organisation of elections, particularly the spatial framework of constituencies and wards. Geographical interest in the location of electoral boundaries can be traced back to Sauer 2. More recently, a number of researchers have focused on the distortion produced by electoral systems in terms of the proportion of votes received and seats won. Two spatial influences have been identified: the location of constituency boundaries and the spatial distribution of voters 3. The assessment of bias has been recorded with particular reference to the plurality system ('first-past-the-post') and its operation in Britain* and New Zealand 5. A critique of the operation of the United Kingdom Boundary Commission has also been developed 6. The discussion has been widened further to consider the extent to which electoral systems frustrate concepts of political justice 7. Although an electoral system functions as an integrated whole, it can, for the purposes of analysis, be broken down into a number of constituent components: (1) franchise regulations such as age or property qualifications; (2) balloting the form for registering a vote, for example the single x-vote or the multiple ranking of choices; (3) districting; (4) the electoral formula. In this paper the districting and electoral formula components are of primary concern. Districting, of particular interest to geographers, has two main components : (a) the division of a political area into units, usually geographical, for the purposes of representation; (b) the magnitude of the district, which can be denned as the number of seats assigned by the electoral law to a district. The electoral formula articulates the translation of votes 42

NORTHERN IRELAND ELECTORAL SYSTEM 43 gained into seats won in each electoral district. The construction of an electoral formula revolves around the problem of how many votes constitute a legitimate claim on a seat. The districting and electoral formula components of an electoral system are closely linked, with the form of electoral formula chosen often determining the appropriate scale of districting. In Northern Ireland, elections and electoral systems have long been surrounded by allegations of manipulation and malpractice. Following the partition of Ireland by the Government of Ireland Act (1920) three levels of political authority were created in Northern Ireland local government, regional parliament (Stormont) and Westminster parliament each based upon mass elections and each with its own electoral system. Allegations concerning the manipulation of the local government electoral system have been given substance by the Cameron Commission 8 and Elliott 9 and some attention has been given to the Westminster system 10. As yet, however, there has been no major assessment of the intervening tier of political authority, that is, the regional parliamentary electoral system. The purpose of this paper is to examine the causes and consequences of the change from a proportional representation electoral system to a plurality system for elections to the regional parliament which took place in 1929. This change in electoral formula and theassociated alterations to the electoral framework is of particular significance as it provoked much opposition from non-unionist groups and has long been alleged to be gerrymandered 11. It is the aim of this paper to assess the motivations prompting the electoral change, to examine the reapportionment undertaken and to consider some of the ramifications of the transition from a proportional to a plurality system. The Government of Ireland Act provided that the initial elections to the regional parliament were to be conducted by means of proportional representation (single transferable vote) using the existing (slightly modified) Westminster constituencies, as multi-number districts returning forty-eight members (plus four from the non-territorial constituency of the Queen's University of Belfast). The electoral system was guaranteed for three years after which the Northern Ireland parliament was empowered to alter all components of the electoral system providing that in any new system the number of members should not be altered and that due regard be given to the population of constituencies other than the university constituency. It was anticipated that the Unionist government would take the earliest opportunity to change to a plurality system. This expectation was based on the opposition to P.R. (S.T.V.) voiced by Unionist MPs at Westminster who had a strong desire to retain the 'British system'. The arguments used by the Unionists against the P.R. (S.T.V.) system were: (1) Most electors were opposed to proportional representation because it was complicated and open to 'manipulation' by agents. (2) It was not generally understood. (3) It involved high costs for candidates. (4) The system produced 'freakish' results. No evidence was presented to back these claims. The urgency of the Unionists, however, was assuaged by the comforting results of the first Northern Ireland general election held in 1921 (a majority of twenty-eight seats over other parties in a fifty-two seat legislature). The results of the 1925 election were less comforting; rigid polarisation of behaviour in evidence in the 1921 election seems to have relaxed somewhat, encouraging Independent and Labour candidates to stand. The returns of 1925 showed a loss of eight seats by the Unionists to Independents and Labour while the anti-partition groups retained twelve seats. Overall, the Unionist majority over other parties fell to twelve seats.

44 OSBORNE This result was regarded if not as a defeat at least as a dangerous trend. The result saw a renewal of the calls for the abolition of P.R. (S.T.V.) by some Unionists. More perceptive Unionists may also have noticed a development in electoral behaviour which was potentially more ominous for the Unionists. In South Antrim and West Belfast, for example. Nationalist votes at the terminal stage 12 showed a tendency to transfer to Labour and Independent Unionists rather than to become non-transferable as had happened in 1921. Such transfers enabled Independent Unionists and Labour candidates to be elected ahead of Unionist candidates. Clearly, if Nationalist voters became prepared to cross the divide and give support to Independents and others the outright majority of the Unionists could have been endangered. In the aftermath of the election then, two new arguments were advanced by the Unionist lobby in their call for the abolition of P.R. (S.T.V.). These were : (1) Strong government was required and not one dependent on bargaining which P.R. was likely to produce. (2) Minority rights were better protected under the plurality system. The Prime Minister, Lord Craigavon, said that what was wanted was, '... a simple old fashioned system where men are for the Union on the one hand or against it and want to go into a Dublin parliament on the other' 13. Quite clearly while a factor in Unionist opposition to P.R. (S.T.V.) was its 'un-britishness' a more pragmatic motivation for abolition was the threat that the P.R. (S.T.V.) system might provide the vehicle for Independents and others to gain a bargaining position through the erosion of the Unionist outright majority. Independents and Labour candidates were seeking support on a broad variety of socio-economic issues; the Unionists sought to base their appeal and strength on the polarisation between pro- and anti-partition viewpoints. P.R. (S.T.V.) threatened to benefit the former rather than the latter. The abolition of P.R. (S.T.V.) took place through the Methods of Voting and Redistribution of Seats (Northern Ireland) Act (1929) which established the plurality system and created forty-eight territorial constituencies. This new system and the spatial pattern of constituencies formed the framework for elections for the forty years to 1969 14. The redistribution of constituencies accompanying the 1929 Act was not undertaken by any independent boundary commission or similar body so there exists no clear outline of operating criteria used. In fact the redistribution seems to have been undertaken personally by Craigavon 15. The basic frame for the redistribution appears to be that of the existing Westminster constituencies which had served as multi-member districts for the 1921 and 1925 elections. There were two exceptions: (i) the separation of Fermanagh and Tyrone (previously a joint eight-seat constituency); (ii) the treating of Belfast as a single unit. The redistribution was governed by the stipulation in the Government of Ireland Act concerning the population of constituencies. The data source used in this respect was the 1926 Northern Ireland population census. The total population for Northern Ireland, taken from the census, was 1,256,651. With a total of forty-eight territorial seats the electoral quota was 26,178. On this basis the redistribution was accomplished as shown in Table 1.

NORTHERN IRELAND ELECTORAL SYSTEM TABLE i THE ALLOCATION OF SEATS BY AREA IN 45 Area Population Population Quota 1929 1921 Antrim 191,642 191,642 26,178 7 + 8,397 7 7 Armagh 110,070 110,070 26,178 4 + 5.358 4 4 Down 209,228 209,228 26,178 7 + 25,982 (8-196) 8 8 Fermanagh 57,984 Tyrone 132,792 57.984 26,178 132,792 26,178 2 + 5,628 5 + 1,902 } 5J 8 Londonderry 139,693 139.693 26,178 5 + 8,803 5 5 Belfast 415*151 15 + 22,481 26,178 (16-3,697) 16 16 The allocation of seats conformed to the pre-existing multi-member allocation of the Government of Ireland Act. In population terms, however, Belfast, Down and Fermanagh were over-represented. The slight overrepresentation of Belfast and Down was not particularly significant as the two areas would have been first in line for the allocation of an additional seat. The allocation of a third seat to Fermanagh when both Antrim and Londonderry were next in line for further seats was more questionable. Craigavon clearly indicated why this had been done : 'The duty that fell to me was to see how, when these two counties that were combined, Fermanagh and Tyrone, were separated again, they would give fair representation of four members to this side of the House and four members to that side' 16. The use of 'fair' was beguiling as the distribution of the additional seat to Fermanagh was designed to achieve a partisan distribution rather than provide a rational distribution of seats on a population basis. Two clearly denned and essentially opposite bases for redistricting either on the basis of population equality principles or to meet a partisan distribution were thus confused. The next step in the redistribution was the allocation of the seats in each area. Craigavon and his colleagues chose to set about this task by calculating a new quota for each area in order to arrive at a target for constituency size rather than using the overall Northern Ireland quota of 26,175. This method,

46 OSBORNE CONSTITUENCIES- 1121 -H I S Figure I. The pattern of constituencies for regional parliamentary elections 1929- I965- given the lack of population parity at the higher spatial level, was bound to increase constituency population inequalities. Moreover, although Belfast had been treated as a single unit for the initial seat allocation, the distribution of seats within Belfast seems to have been enacted in terms of the existing configuration of Westminster constituencies; this produced marked disparities in the seat allocations within the city as shown in Table 2. TABLE 2 THE ALLOCATION OF SEATS IN BELFAST IN 1929 Constituency Belfast East Belfast North Belfast South Belfast West 3 3 3 3 Quota + I5,78i + 23,076 + 13.177 + 25,806 Allocation 4 4 4 4 The pattern of the forty-eight new constituencies can be seen in Figure 1. The overall basis of the redistricting can be criticised in a number of respects: first, the decision to adhere to the existing seat allocations of the

NORTHERN IRELAND ELECTORAL SYSTEM 47 Westminster constituencies rather than derive an allocation based on the principle of equality of population per seat; second, the confusion of two clearly defined and opposite bases for redistricting, i.e. population equality or maximising a desired distribution; finally, the decision to distribute the seats within each base area in terms of a quota derived from that area. A more direct critical assessment, however, can be made by examining the redistricting for malapportionment and gerrymandering. Malapportionment involves the creation of constituencies of markedly different sizes in terms of either population or electorate. Malapportionment is held to be 'bad' as it implies that the value of a vote is less in a larger constituency than in a smaller one. The nearer that constituencies can be made to equal each other in electorate/population size the greater is the equality of vote value. The upsurge of interest in the districting problem in the U.S.A., Australia and Great Britain has led to the development of a number of techniques which can be applied to test statistically for malapportionment and which facilitate comparative analysis. Alker and Russett have suggested two tests as particularly useful for measuring inequality in population distribution. These are the Lorenz curve and the associated Gini Index and the Inverse Coefficient of Variation (I.C.V.) 17. These indices have been calculated for the populations of the new constituencies (1926 population census) and the electorates at the first election of 1929. They have been calculated for the forty-eight constituencies in the first instance and then for Belfast (Table 3). TABLE 3 GINI AND I.C.V. INDICES FOR CONSTITUENCY ELECTORATES AND POPULATIONS Index Northern Ireland 1926 Population 1929 Electorate 1926 Population Belfast 1929 Electorate Gini.11 07.10.09 I.C.V..89.88 85.85 In Table 4 comparative figures are given (I.C.V.) for three of the U.K. regional areas including Boundary Commission recommendations. From this evidence, the 1929 redistricting in Northern Ireland does not show any evidence of malapportionment with a Gini score of 0.11 and I.C.V. score of 0.89. The figures are similar when calculated for constituency electorates. If these Northern Ireland figures are compared with the U.K. figures the absence of malapportionment seems to be confirmed. For example, in relation to Wales (thirty-six constituencies) the Northern Ireland I.C.V. of 0.88 (electorate) compares favourably with a score of 0.82 achieved with both 1965 and 1968 Boundary Commission proposals. Perhaps, however, the absence of large-scale malapportionment in the 1929 redistribution masks the extent to which constituency parity was not achieved. The comparison with Wales, for example, may be over-favourable to Northern Ireland in that in Northern Ireland a completely new framework for territorial constituencies

48 OSBORNE was being devised whereas in Wales an existing framework was under adjustment and certain criteria-constraints prevented a more equal apportionment being achieved. Certainly when the figures for Belfast are separated out they seem to indicate that a less equal apportionment was achieved than when the province as a whole is considered. However, overall it must be concluded that there is no evidence of large scale malapportionment associated with the 1929 redistricting. Qualifications must be recorded, however, in that disparities in population size have been noted and that these can be attributed to several decisions made by Craigavon and his colleagues. TABLE 4 I.C.V. SCORES FOR BRITISH REGIONS IN 1965 AND 1968 (INCLUDING BOUNDARY COMMISSION RECOMMENDATIONS) 1965 1968 Region Actual Recommended Actual Recommended England.84..88.82.87 Scotland.81.84.78.84 Wales.82.82.82.82 U.K..82.86.80.85 If the spatial framework established in 1929 cannot be described as malapportioned it still remains to be examined for gerrymandering. Essentially there are three means of achieving a gerrymander. In the first method opposition votes can be wasted by designing constituencies where they poll not more than 49 per cent of the vote. In such cases opposition electoral strongholds can be divided between a number of constituencies so as to maintain a consistent voting minority. The second method is by enclosing opposition votes in as few constituencies as possible. These constituencies are won with massive majorities while the other party wins more seats with narrow majorities. The third method consists of gathering together scattered areas of support in curiously shaped constituencies so that seats may be won. It is apparent that these methods are not always easily identified and careful scrutiny is a priority. Taylor 18 has argued that a non-partisan cartographical division of areas is mythical and so following this logic it might be argued that gerrymandering is unimportant as it merely represents another source of bias. Gerrymandering, however, implies a motivation on the part of the constituency cartographer which may well be politically more significant than the results it actually achieves as it demonstrates a degree of cynicism in the political system which is bound to have repercussions in the political community as a whole. A full analysis, covering gerrymandering, of the spatial framework of the 1929 reorganisation can be found in Osborne. 19 Three main areas of concern emerged from this examination. The first concerns the allocation of a third seat to Fermanagh. This award was made notwithstanding the prior claims of Antrim and Londonderry (Table 1). Craigavon's desire to maintain a

NORTHERN IRELAND ELECTORAL SYSTEM 49 PERCENTAGE ROMAN CATHOLICS BY D.E.D., Co.FERMANAGH (1911 CENSUS) N. IRELAND QUOTA (Popubtion) - 36,17«Cwwtituancy Enniakillwi L<tna*k«a South Population HIM 1 Î.5 6 3 1 9,3 0 7 1 P.l 1 2 Elector*«13.33? 11.6 6t 11,4 12 KEY * County boundary - Conatitutrtcy boundary - D.E.D. botindary % Roman Catholic* Figure 2. Constituencies and the distribution of Roman Catholics in Co. Fermanagh (1911 Irish census). partisan seat distribution has been shown to underlie this decision. The allocation of seats for Tyrone was reasonable five quotas and five seats. The partisan distribution of these seats was acknowledged to be three Nationalists and two Unionists. So, given that Craigavon wished to preserve the 1921 and 1925 distribution of seats in the then combined Fermanagh- Tyrone constituency of four Unionists and four Nationalists the task was to ensure that the three Fermanagh constituencies produced two Unionists and one Nationalist. How this was achieved can be seen in Figure 2. Here the county of Fermanagh is displayed with the three constituency boundaries. Also shown are the District Electoral Divisions (D.E.Ds) which are shaded to depict the percentage Roman Catholic in each area, derived from the 1911 Irish census 20. From this map it can be seen that the Fermanagh South constituency was created by linking the majority Catholic D.E.Ds along the southern border. The constituencies of Lisnaskea and Enniskillen were then created by splitting the remainder of the county in half on a north-south axis and dividing the remaining areas with a Catholic majority between the two, rendering them a minority in each. Subsequently Fermanagh South was held by the Nationalists in the period 1929-1969 and Enniskillen and Lisnaskea by the Unionists. Indeed, the patterns became so well known locally that only in 1949 were all three seats contested by the Unionists and Nationalists. As can be seen from Table 5 the two Unionist seats contained a substantial opposition vote. This pattern fits one of the gerrymander 'models' outlined above, with Nationalist votes gathered in Fermanagh South and the remainder 'wasted' in seats where they polled highly but were never

50 0SB0RNE sufficient to win the seat. In fact both parties recognised the situation with all three seats normally returned unopposed. TABLE 5 CONSTITUENCY RESULTS FOR FERMANAGH IN 1949 Constituency % Unionist % Nationalist % Turnout Fermanagh South Enniskillen Lisnaskea 28.0 54-7 57-3 72.0 45-3 42.7 86.6 85.7 87.2 Source: S. Elliott, Northern Ireland parliamentary election results 1921-1972, Chichester, 1973. In this way, by using detailed local information as a basis for partisan electoral cartography, Craigavon was able to produce his desired seat distribution in a county where Catholics formed 56% of the total population and 53% of the population of voting age. The second major area for examination was Antrim where the Nationalists were vociferous in their dislike of the configuration of constituencies. The allegation here was that whereas in the 1921 and 1925 elections a Nationalist had been returned this was unlikely to happen with the new distribution. Once again the basic framework for examining this situation is the religious distribution in the county. In Table 6 the numbers and proportions of Catholics are shown for the various urban and rural districts (1926 census). These figures demonstrate that critically, Ballycastle, the only urban district with a Catholic majority, was the smallest of the eight urban districts in County Antrim. The largest numbers of Catholics in urban districts were found in Ballymena, Larne and Lisburn where they in fact formed small minorities. Again in the rural districts, the highest percentage of Catholics was in Ballycastle which however was the smallest rural district in population terms. The Ballycastle urban and rural districts together formed the only majority Catholic area in Antrim. It was far too small, however, to form a constituency in its own right (population 11,433; Antrim quota 27,378; smallest constituency in Northern Ireland 18,000). Nationalists were adamant in their insistence that they were being deprived of a seat and pointed to Craigavon's wish to preserve the pre-1929 partisan seat distribution in Fermanagh-Tyrone why could this not apply to Antrim? Clearly, the only way this could be done was either to create a very small constituency around Ballycastle or to link other small majority Catholic areas, thereby relaxing the contiguity-of-areas principle. These ideas were rejected by the Craigavon administration 21. The third main area of assessment for gerrymandering is County Armagh. The distribution of Catholics is once more the critical base from which to examine for gerrymandering. In Figure 3 the percentages of Catholics are shown for D.E.Ds from the 1911 Irish census. From this map it can be seen that Catholics are clustered in the southern third of the county with only a few small groups of D.E.Ds where Catholics are a majority outside this area.

NORTHERN IRELAND ELECTORAL SYSTEM TABLE 6 DISTRIBUTION OF ROMAN CATHOLICS IN Co. ANTRIM (1926 Northern Ireland Population Census) Urban Districts Total Population Catholics % Catholic Ballycastle 1,986 1,135 57.2 Ballyclare 3,362 294 8.7 Ballymena 11,873 2,154 18.1 Ballymoney 3,168 723 22.8 Carrickfergus 4.749 691 14.6 Lame 9,714 2,248 23.1 Lisburn 12406 2,194 17.7 Portrush 2,953 449 15.2 Rural Districts Antrim 25.112 4 8 17.1 Ballycastle 9.447 4,452 47.1 Ballymena 33.257 5,606 16.9 Ballymoney 20,305 3,978 19.6 Belfast 13.857 3,134 22.6 Larne 22,089 3>548 16.1 Lisburn 17,355 3,73 21.5 Given this distribution the orientation of constituency boundaries is clearly critical in deciding the ultimate partisan allocation of the four single-member constituencies. The shape of Armagh is generally rectangular and with this shape there are at least two distinct ways of dividing the area into four. One is by means of a series of broadly parallel east-west lines as the 1929 distribution was accomplished and the other is by quartering the area by means of à north-south line crossed by an east-west line. Choosing the former allows the majority Catholic area to be contained mostly in one unit. By choosing the latter it is at least feasible that the Catholic areas would form a majority in the two southern quarters. The patterns of representation for the two elections prior to the 1929 redistricting show that for both 1921 and 1925 the Unionists won two seats and anti-partitionists also two. From 1929 onwards the partisan distribution was Unionist three seats and anti-partition groups one seat. On this evidence there appears to be a strong case supporting the view that the Armagh constituency boundaries were gerrymandered. However a closer scrutiny of the voting returns in the 1921 and 1925 elections suggests that in this case the assertion must be qualified. In 1921 the fourth

52- OSBORNE PERCENTAGE ROMAN CATHOLICS BY D.E.D., Co. ARMAGH (1)11 CENSUS) N.IRELANC> QUOTA (Population)- 2S.I7«Cenatituancy Population Electoral* North Cantr. Mid 26,3 21 1 ö.3 6a 37,86» 16.72» 3 6.8 0 0 16,3 61 KEY County boundary. Conatituancy boundary LU] 5533 DEC boundary Roman Catholic««15 36-4» SO-44»15 Figure 3. Constituencies and the distribution of Roman Catholics in Co. Armagh (1911 Irish census). Armagh seat was won by a Nationalist/Republican candidate who failed to attain the quota and in 1925 the third and fourth seats were taken by antipartition candidates without the votes of the runner-up being redistributed. Even if these votes had been redistributed, they would have been insufficient to enable the last elected to attain the quota. Added to this is the fact that with" the Unionists only nominating two candidates many of their surplus votes became non-transferable after election was achieved. In 1925, for example, the number of non-transferable votes from the two Unionists amounted to more than an additional quota, which may have produced a further seat for the Unionists, given a further candidate. Thus, in relation to the Armagh constituency boundaries the case for gerrymandering is not conclusive. While anti-partition representation for the county was reduced by one the pre-1929 representation may have been flattering to the Nationalists through the poor nomination tactics of the Unionists. In these examples the importance of two spatial components, the spatial distribution of votes and the configuration of constituencies, can be seen to have been of crucial importance. The effects resulting from the different levels of vote aggregation imposed by different electoral.formulae (P.R. (S.T.V.) and plurality) can be established. In the case of Antrim for example, the scale of aggregation (linked with arguments concerning electoral quotas and the number of available seats) determined the non-representation of the politically distinctive social group concentrated around Ballycastle which

NORTHERN IRELAND ELECTORAL SYSTEM 53 was too small to secure representation and so for Stormont elections became submerged under the majority grouping. Overall, the 1929 redistribution was accomplished without evidence of malapportionment. Gerrymandering was undoubtedly a feature of the Fermanagh seat distribution and may also have been involved in Armagh. However, on the basis of this evidence it is hardly possible to call the 1929 redistribution a general exercise in gerrymandering 22. Also, given the 'myth of non-partisan cartography' it becomes clear that the degree of partisan redistricting shown here provides rather slender evidence and scarcely matches the volume of Nationalist opposition which followed in its wake 23. Nevertheless, the existence of some gerrymandering at this level coupled with the evidence from local government redistribution was unlikely to increase Nationalist commitment to the new system. The final part of this assessment of the 1929 changes concerns the electoral formula component of the electoral system. Some of the implications of the transition from a P.R. (S.T.V.) to a plurality formula have been seen already. One further aspect which requires assessment is the measurement of the 'bias' in the system. If one of the foundations of representative democracy is the relationship between votes and seats 2 *, then the extent to which an electoral system distorts the translation of votes and seats is of concern. Loosemore and Hanby 25 have outlined an index suitable for an assessment of distortion. Using this index the scores are shown for the 1921-1938 elections in Table 7. It is readily apparent that distortion in the votes-seats relationship increased markedly with the transition from P.R. (S.T.V.) to plurality. TABLE 7 DISTORTION IN THE VOTES-SEATS RELATIONSHIP, NORTHERN IRELAND PARLIAMENTARY ELECTIONS 1921-1938 Election 1921 1925 1929 1933 1938 Distortion score.101.083.301.302.293 This index, however, does not take account of the influence of uncontested seats which play a considerable role in Northern Ireland's electoral history (see below). Although distortion increased quite dramatically following the adoption of a plurality system, it is important to consider which groups benefited and which lost through this increase. In Table 8 an indication of the distribution of gain and loss is shown for the major groupings. In 1925, the two smaller groupings, the N.I.L.P. and the Independent Unionists, benefited while the Nationalistsand Unionists failed to gain seats in proportion to their votes.

54 OSBORNE TABLE 8 PARTY GAIN AND LOSS IN THE VOTES-SEATS RELATIONSHIP, NORTHERN IRELAND PARLIAMENTARY ELECTIONS 1925-1938 (Calculated from figures given in Elliott, op. cit., 1973) PanyjGrouping 1925 1929 1933 1938 Unionist + + + Independent Unionist + NILP + + Nationalist + + + In the next three elections using the new system, however, the pattern of gain and loss is reversed with the larger parties, the Nationalists and Unionists, profiting. This distribution of bias in the votes-seats relationship with the larger parties benefiting at the expense of smaller parties is the pattern found in most countries having a plurality electoral system. The political implications and consequences of this alteration of the electoral system can now be briefly considered. The Unionist motivation for change clearly lay with the desire to eliminate the smaller parties and independents and to sidetrack attempts to focus attention on broad socio-economic issues. It can be suggested, therefore, that the 1929 redistribution was accomplished with partisan ends in mind but the target was not so much the opposition Nationalist parties as the minority parties and independents who threatened to gain from the P.R. (S.T.V.) system and seriously erode the Unionist party's outright majority. The Nationalist party consistently profited in terms of seats won in relation to the proportion of votes gained and their representation showed only a loss of one or two seats over the earlier totals. After 1929, splinter parties failed to win the seats their overall totals warranted. They suffered from increased logistical problems because of the rise in territorial constituencies from nine to forty-eight. While the 1929 redistricting cannot be characterised as malapportioned or the spatial framework of constituencies as subject to wholesale gerrymandering, the manipulation of the electoral system had more serious implications. The reduction in thé spatial scale of electoral districts coralled electors into neat areas of Nationalist or Unionist dominance which became tacitly accepted as such. This acceptance of areas of partisan dominance was reflected in the large numbers of uncontested seats which only once fell below 40% of the forty-eight territorial constituencies in the nine general elections in the 1929-1965 period. Electoral competition was often generated only by the activities of individuals and the smaller parties who sporadically challenged the hegemonies of the main parties. This intervention showed little spatial or temporal continuity other than in several Belfast constituencies. Apart from representing a parody of electoral competition in a democratic system, the firm demarca-

NORTHERN IRELAND ELECTORAL SYSTEM 55 tion and stability of the areas showing governing party loyalty provided the spatial context for decisions concerning the allocation of public resources and investment. In the post-war period, decisions concerning the siting of the province's second university at Coleraine, the location of the new city of Craigavon and the spatial pattern of growth centres for regional development were increasingly criticised as being made with this pattern of support in mind 26. However, this is judging the new 1929 electoral system beyond the 'fairness' of the actual framework devised and it is clear that the Northern Ireland government would not be unique in making vote-catching locational decisions. Aside from this however, the politico-geographical implications of the 1929 reorganisation were far reaching. The establishment of a plurality system and its spatial framework performed exactly the task desired by Craigavon. By reducing the spatial scale and increasing the number of electoral districts the smaller parties, who sought to offer socio-economic issues to the electorate, were inhibited from making a geographically widespread appeal and by structuring electoral choice to a single no-compromise x-vote the potential for cross-voting was removed. The new electoral system thus provided a structural prop to the communal divisions of Northern Ireland and produced an electoral pattern which neatly fulfilled the requirement to show 'where men are for the Union on the one hand or against it and want to go into a Dublin parliament on the other'. NOTES AND REFERENCES 1 I. R. McPhail, 'Recent trends in electoral geography', Proceedings Sixth New Zealand Geographical Conference, Christchurch, 1970, 7-12. M. Busteed, Geography and voting behaviour, London, 1975. P. J. Taylor, 'Political geography'. Progress in human geography, 2, 1978, 153-162. 2 C. Sauer, 'Geography and the gerrymander', American Political Science Review, 12, 1918, 403-426. 3 P. J. Taylor, 'Some implications of the spatial organization of elections'. Transactions, Institute of British Geographers, 60, 1973, 121-136. 4 G. Gudgin and P. J. Taylor, 'Electoral bias and the distribution of party voters', Transactions, Institute of British Geographers, 63, 1974, 53-74. 5 R. J. Johnston, 'Spatial structure, plurality systems and electoral bias', Canadian Geographer, 20, 1976, 310-328. 6 G. Rowley, 'Elections and population changes', Area, 3, 1970, 13-18. G. Rowley, 'The redistribution of parliamentary seats in the United Kingdom: themes and opinions', Area, 7, 1976, 16-21. 7 R. J. Johnston, 'Parliamentary seat redistribution: more opinions on the same theme', Area, 8, 1976, 30-34. 8 Disturbances in Northern Ireland. Report of the commission appointed by the Governor of Northern Ireland (Cameron Commission), HMSO, 1969. 9 S. Elliott, The Northern Ireland electoral system since 1920, Unpublished Ph.D. thesis, Queen's University, Belfast, 1971. 10 R. D. Osborne, 'Northern Ireland: representation at Westminster and the Boundary Commission', Ir. Geog., 10, 1976, 115-120. 11 This spatial framework no longer exists having been replaced by the twelve Westminster constituencies assigned multi-member magnitude for elections to any regional political authority under the Northern Ireland Constitution Act (1973). 12 An essential element of P.R. (S.T.V.) is the transfer of vote values. These can be classified into two types. The first is the interim transfer which takes place when there are still members of the transferring candidate's party available to receive transfers. The second is the terminal transfer which takes place when there are no members of the candidate's party available to receive vote values, all having been either elected or eliminated.

56 OSBORNE 13 House of Commons Debates (NI), 25.10.27, 2275. 14 The university constituency was abolished following the recommendations of an Interim Boundary Commission (1967) which created four new territorial constituencies. One general election was fought on this framework in 1969. 15 'I have taken great pains over it. I have taken months in order to ensure...' (House of Commons Debates (NI), 10, 26.3.29, 1450). 16 House of Commons Debates, (NI) 10, 26.3.29, 1450. 17 H. R. Alker and B. Russett, 'On measuring inequality', Behavioural Science, 9, 1964, 207-218. The Gini Index indicates the difference between the ideal and actual relationship with lower Gini scores indicating more equal apportionment. The Inverse Coefficient of Variation produces a standardised measure of the variance inverted so that it ranges between o and 1. Low values indicate malapportionment and a value of 1 indicates perfect apportionment. 18 P. J. Taylor, 'Some implications of the spatial organization of elections', Transactions, Institute of British Geographers, 60, 1973, 121-136. 19 R. D. Osborne, The political system, voting patterns and voting behaviour in Northern Ireland, 1921-1974, Unpublished Ph.D. thesis, Queen's University, Belfast, 1977. 20 The 1911 population census was the last Irish census to publish information for religious affiliation at this scale. Percentages are shown rather than exact totals, as the D.E.D. boundaries were altered between the census and the establishment of constituency boundaries in 1929. 21 Even if Antrim had been allocated the additional seat awarded to Fermanagh it would still have been difficult to create a constituency around Ballycastle in terms of population size. If, however, the aim had been to preserve the earlier pattern of representation, as in Fermanagh, such difficulties would not have been insurmountable. 22 Taking the simple comparison of constituency electorate size between Unionist and Nationalist seats (in rural areas) shows the average Unionist seat with 16,547 electors compared to 15,406 for the Nationalist constituencies in 1929. 23 Labour activists were quite aware of the likely impact on their representation after reapportionment and sternly opposed the change from P.R. (S.T.V.). When Nationalists argued for relaxing the contiguity-of-areas principle in Antrim, Labour immediately demanded a seat allocation for the working class in Belfast. 24 J. G. March, 'Party legislative representation as a function of election results', Public Opinion Quarterly, 21, 1957, 521-542. 25 J. Loosemore and V. J. Hanby, 'The theoretical limits of maximum distortion: some analytic expressions for electoral systems', British Journal of Political Science, 1,19 1971, 467-477. 26 See the debates of these issues in the Northern Ireland House of Commons Debates, volumes 55-61, covering the 1964-1965 period. ACKNOWLEDGMENTS The author would like to thank Dr Neville Douglas, Department of Geography, Queen's University, Belfast, for his helpful comments on an initial draft of this paper and Merle Osborne for assisting with the drawing of the maps.