ANNALS OF HUMAN BIOLOGY, 1983, VOL. 10, NO. 4, 321--334

Genetic structure and population history of Ireland: a comparison of blood group and anthropometric analyses

J. H. RELETHFORD

Department of Anthropology, State University of New York College at Oneonta, NY

Received 19 January 1983

Summary. Population structure and history may be studied on a local or a regional level. This paper examines the regional population structure of the Republic of Ireland and Northern Ireland with respect to population history and demographic processes. Blood- group and anthropometric data obtained from the literature are analysed. The blood-group data consist of ABO and Rhesus gene frequencies for 32 counties and the Aran Islands. Anthropometric data consist of summary statistics for 15 variables collected from 19 regions. The degree and pattern of population differentiation is assessed using new methods of population-structure analysis. Both blood group and anthropometric analyses show a west-east division of populations corresponding to the known history of inhabitation of Ireland, where successive waves of immigrants pushed earlier populations further west. In both analyses there were two deviations to this basic pattern: the Aran Islands and the mid- lands. In both cases, alternative historical explanations are examined. The genetic relationship of the Aran Islands to the rest of Ireland and England appears to be due to English admixture following the garrisoning of soldiers several centuries ago. The genetic position of the midlands is more complex, but suggests the effects of early Viking inhabitation. These findings are related to studies of the local, rather than regional, popu- lation structure of Ireland.

I. Introduction In recent years, studies o f human popu la t ion structure have al lowed the

investigation o f cul tural and biological in terre la t ionships in terms o f the analysis o f genetic effects o f mat ing structure. Mat ing probabi l i t ies can be influenced by the composi t ion o f a popula t ion , preferences a n d / o r avoidances o f consanguineous marr iage, and barr iers to gene-flow, such as geographic dis tance and social class. His tor ical ly or iented studies have al lowed changing pat terns o f popu la t ion structure to be assessed. Here, known migra t ion pat terns , popu la t ion movements , changes in popula t ion size, and general aspects o f demograph ic change are shown to affect the genetic structure o f regional popula t ions . Studies o f regional , ra ther than local , popula t ion structure have been done for a variety o f Eu ropean countries, including Belgium (Dodinval 1970), Switzer land (Mor ton , Klein, Hussels, Dodinval , Todorov, Lew and Yee 1973), F in land (Workman , Mielke and Nevanl inna 1976), I ta ly (Zanardi , Del l 'Acqua, Menini and Barra i 1977), Iceland (Jorde , Er icksson, M o r g a n and W o r k m a n 1982), and England (Roberts , Jo rde and Mitchell 1982).

I re land (including the Republ ic o f I re land and Nor the rn I re land) is a count ry whose unique his tory and demography have long interested social scientists (Kennedy 1973). To date, research on genetic structure and popu la t ion his tory o f I re land have fol lowed two different paths. Firs t , there have been a number o f large regional studies focusing on basic pat terns of var ia t ion with respect to his tory on a t ra i t -by- t ra i t basis. These studies have dealt with bo th an thropomet r ics ( H o o t o n and Duper tu is 1955) and b lood groups (Hacket t , Dawson and Dawson 1956, Hacke t t and Dawson 1958, Dawson 1964, Tills 1977 a, Tills, Teesdale and M o u r a n t 1977). These studies have been

general ly descriptive, univar ia te , and not in tegra ted in the context o f popu l a t i on structure theory. The second type o f s tudy has dealt with local popu la t ion structure,

322 J. H. Relethford

focusing on rural populations in western Ireland (Relethford 1980 b, Relethford, Lees and Crawford 1980, 1981, Lees and Relethford 1982) and west-coast isolates (Relethford 1982). These studies have dealt with the methods and theories of population structure, but describe only those patterns of interest on a local level. It is expected that the cultural and demographic processes operating on genetic structure will be different on a regional level.

The purpose of this paper is to investigate the regional population structure of Ireland using newly developed methods of population structure analysis. The main emphasis of this analysis is to relate genetic structure to population history. Results are compared with other European populations and with studies of local population structure in Ireland.

2. Materials Blood-group and anthropometric data are available in the published literature

for a regional analysis of Ireland (The term 'Ireland' as used in this paper refers to the Republic of Ireland and Northern Ireland collectively). Blood-group data are limited to the ABO and Rhesus loci. While several studies have incorporated additional loci, they are limited to only a few populations or the entire island as a whole (e.g., Sunderland, Tills, Bouloux and Doyle 1973, Walter and Palsson 1973, Tills 1977 a, Tills et al. 1977). In order to include a large number of populations representing the entire region, all blood-group analyses had to be confined to the two loci. While a large number of additional loci would be preferred, the use of only the two loci is still highly valuable, as shown by a number of studies on population structure where only these loci (or even just one) were available (e.g. Imaizumi and Morton 1969, Dodinval 1970, Morton et al. 1973, Zanardi et aL 1977, Jorde et al. 1982). In any case, the usefulness of any data set rests on the ability to interpret the results in a meaningful manner.

The basic unit of analysis for blood-group study in this paper is the county. The Republic of Ireland has 26 counties and Northern Ireland has 6. In addition to these 32 counties, the Aran Islands have been treated as a separate population. The Aran Islands consist of three small islands in Galway Bay off the west coast of County Clare and County Galway. The islands have been treated as a separate population due to their geographic isolation and previous reports of suggested genetic divergence from the mainland of Ireland (Hooton and Dupertuis 1955, Hackett and Folan 1958). The locations of all the counties are shown in figure 1. Three studies were used to provide the gene frequencies needed for analysis. Data for the 26 counties in the Republic of Ireland were taken from Dawson (1964), which lists data for 117 287 adult men and women. This sample is approximately 5-5°70 of the total adult population of the Republic of Ireland at the time of sampling. Gene frequencies for the Aran islands were taken from Hackett and Folan's (1958) study of 229 people. Gene frequencies for the six counties in Northern Ireland were taken from Tills et al. (1977), who compiled the earlier data of Kope6 (1970). This sample consists of 27 861 men and women for the ABO locus and 32799 men and women for the Rhesus locus. Overall, the sample sizes from these studies are high, compensating to some extent for the limited number of loci for analysis.

Anthropometric data are also available on a regional basis. These data were taken from Hooton and Dupertuis (1955), and were collected originally as part of

Genetic structure o f Ireland 323

Figure 1. Location of Populations used in the blood-group analysis.

1, Aran Islands; 2, Mayo; 3, Galway; 4, Clare; 5, Limerick; 6, Kerry; 7, Sligo; 8, Roscommon; 9, Tipperary; 10, Cork; 11, Donegal; 12, Leitrim; 13, Longford; 14, Westmeath; 15, Offaly; 16, Leix; 17 Kilkenny; 18, Waterford; 19, Cavan; 20, Meath; 21, Kildare; 22, Carlow; 23, Monaghan; 24, Louth; 25, Dublin; 26, Wicklow; 27, Wexford; 28, Fermanagh; 29, Tyrone; 30, Derry; 31, Armagh; 32, Antrim; 33, Down. The Aran Islands (1) are part of County Galway (3) but are treated separately here. Counties numbered 28-33 are in Northern Ireland.

the Harvard Anthropological Survey of Ireland in the 1930s. The data are reported in 'county subgroups' , which were defined by Hooton and Dupertuis as single counties, multiple counties, or subgroups composed of parts of various counties. Nineteen county subgroups were used by Hooton and Dupertuis, and are shown in figure 2. Since the units of analysis are not the same for blood-groups and anthropo- metric data, certain analytical techniques, such as correlation analysis and rotational fitting, could not be used, and comparison is limited to visual inspection of patterns of variation. While a portion of the original data have been analysed in previous papers (Relethford 1980 a, b, Relethford, Lees and Crawford 1980, Relethford and Lees 1981, Relethford et al. 1981, Lees and Relethford 1982, Relethford and Lees 1983), only summary data are available for the entire population of Ireland. Also, these summary data are limited to adult males, since data on adult females were collected only in a few western counties. The summary data consists of means and standard deviations for 15 anthropometric variables: stature, biacromial breadth, chest breadth, chest depth, sitting height, head length, head breadth, head circumference, head height, bizygo- matic breadth, bigonial breadth, minimum frontal diameter, upper facial height, nose length, and nose breadth. Since environmental variation may contribute to among group variation in body size, all analyses were repeated using only the ten craniofacial variables, since they are often quite sensitive in detecting genetic affinities (Fried- laender 1975, Relethford et al. 1980).

324 J.H. Relethford

Figure 2. Location of populations used in the anthropometric analysis.

1, Aran Islands; 2, Mayo; 3, West Galway; 4, Clare; 5, Kerry; 6, West Donegal; 7, Sligo, Leitrim, South Fermanagh, West Cavan; 8, East Galway, Roscommon; 9, Limerick; 10, Cork; 11, Lonford, Westmeath; 12, Offaly, Leix, Kildare; 13, Tipperary, Kilkenny; 14, Waterford, Wexford; 15, Meath, Lough, Dublin; 16, Wicklow, Carlow; 17, North Fermanagh, East Donegal, Tyrone, Derry; 18, East Cavan, Monaghan, Armagh; 19, Antrim, Down.

3. Methods Population affinities were assessed using measures of genetic structure and

microdifferentiation. For blood-group gene-frequency analyses, the R matrix technique of Harpending and Jenkins (1973) was used. Here, a matrix of standardized genetic variances and covariances among populations is computed for the kth allele as

(Pik -- p k ) (Pjk -- p k) ruk = (1)

Pk (1 - Pk )

where Pik and Pie are the gene frequencies for allele k in populations i and j , and Pk is the weighted mean gene frequency for allele k over all populations. Each element in the R matrix is then computed as the mean value over all alleles. Each element in the matrix represents divergence or co-divergence from the contemporary regional gene-pool, and has also been referred to as 'conditional kinship' (Morton, Yee, Harris and Lew 1971). Positive rii values indicate a pair of populations more similar

Genet ic s t ruc ture o f I r e land 325

than expected at random, while a negative rij value indicates the opposite. A mean value of conditional kinship within groups, RST, may be computed as the weighted mean of all rii values. Additionally, this value may be computed after adjustment for sampling bias (Workman, Harpending, Lalouel, Lynch, Niswander and Singleton, 1973). The RsT value is an estimate of the average divergence from a contemporary gene-pool. As such, RST is similar to Wright's Fsx , except that the latter refers to divergence from a founding, and not contemporary, gene-pool.

Since visual inspection of a 33 by 33 matrix of conditional kinship values precludes any comprehensive synthesis, this information must be represented in a more usable manner. Here, the method of Harpending and Jenkins (1973) is used, where the eigenvalues and eigenvectors of the R matrix are extracted using principal components analysis. The eigenvectors represent the positions of the populations along scaled axes which have been extracted in descending order of explained variation. As such, often only a few axes are needed to provided an accurate representation of genetic variation in a multidimensional space. Patterns of variation among alleles can also be investigated in order to detect which alleles contribute the most (or least) to specific patterns of among-group variation (Harpending and Jenkins 1973).

Anthropometric variation is also assessed using a measure of population differentiation. Here, the Euclidean distance between each pair of populations is computed as

d 2 = ~ (Xi~ - XJk) : (2)

k

where gik and "~jk are the standardized means for variable k for populations i andj . Euclidean distance is directly related to conditional kinship as shown by Morton et al. (1971) and Relethford (1980 a). Thus, Euclidean distance can be used for genetic inference assuming that environmental and sampling effects are either minor or distributed at random across all populations. A measure such as Mahalanobis' D 2 would be preferred since it controls for intercorrelation and redundancy of information, but is not possible to use given summary data. An eigenvectorial plot can be produced from the distance matrix using Gower's (1966) method of converting distances into similarity measures and then using the standard principal components method.

4. Results The average amount of divergence from the regional gene-pool (RsT) is 0" 00113

for the ABO system and 0- 00057 for the Rh system, giving an overall value averaged over five alleles of RST = 0" 00091. After correction for sampling bias, the estimates are similar: ABO = 0.00102, Rh = 0.00047, total = 0.00080. The overall level of genetic microdifferentiation is greater than values for regional studies in Scotland (0-00030), England (0.00034) and Czechoslovakia (0.00031), similar to values from France (0.00087) and Sweden (0.00078), and less than values from Finland (0.00220) and Ireland (0.00123) (comparative values were taken from Jorde 1980). Overall, the Rsv value for Ireland is in the range expected. The values reported in the present study are lower than those given by Tills (1977 b) in his analysis of gene-frequency variation among Irish counties (ABO = 0.01200, Rh = 0.01875, combined -- 0.01538). In that

326 J. H. Relethford

study, however, the sample sizes were extremely small, with a total sample of 1793 for the ABO locus and 1 896 for the Rh locus (Tills et al. 1977).

While the average value (RsT) is low, some of the individual counties show great divergence from the regional mean. The three largest values of rii are for County Clare (0.00814), the Aran Islands (0.00691) and County Cork (0.00403). By contrast, the three lowest values are for County Meath (0.00007), County Derry (0" 00011) and County Kildare (0" 00012).

Individual variation in genetic divergence can be shown more clearly by an examination of the eigenstructure of the R matrix. Figure 3 is a plot of the first two eigenvectors extracted from the R matrix, with eigenvector 1 accounting for 69" 5% of the total variation, and eigenvector 2 accounting for 17.2%. Each axis in figure 3 has been scaled by the square root of its corresponding eigenvalue in order to provide a more accurate picture of relative dispersion (Harpending and Jenkins 1973).

• 23

• 11 • 24

0 2 9 • 5 016

07 021

31 i

°06 02 12 o o o:~ 4 10

20 30 015

9

• 18 026

0 8 • 13

Figure 3. Plot of the first two eigenvectors of the R matrix based on blood groups. Each axis is scaled by the square root of the corresponding eigenvalue.

The first axis separates populations along a west-east division for the most part. The populations grouping to the extreme left are those located on or adjacent to the west coast of Ireland: Clare (4), Cork (10), Roscommon (8), Galway (3), Kerry (6) and Mayo (2). Those populations plotting to the immediate left of centre are those in the midlands. For the most part, populations plotting to the right of centre are those which are on, or adjacent to, the east coast. Several of these populations are markedly separated from the rest, including Armagh (31), Wexford (27), Wicklow (26) and Fermanagh (28). Finally, the Aran Islands (1) plot to the extreme right, in total opposition to the general west-east pattern. Taken as a whole, the Northern Irish populations (28-33) tend to plot further to the right than their neighbours in the Republic.

Overall, the west-east pattern shown in figure 3 could have two interpretations. First, that geographic distance acts to subdivide the entire population according to the isolation by distance model, such that the further two populations are in geographic space, the more genetically dissimilar they will be (Jorde 1980, among

Genetic structure o f Ireland 327

others). The second interpretation is that repeated waves of immigrants in the past have pushed previous populations further westward (Dawson 1964). The separation of populations along the second axis weakens the isolation-by-distance argument, at least as applied to isotropic migration. Instead of a north-south separation, the second axis acts to separate the populations of Roscommon (8) and Longford (13). The third axis (not shown) accounts for the remaining 13-2070 of the total variation, and acts only to separate County Clare (4) even more. Plots of variation among alleles (not shown) show that the basic west-east pattern represents a contrast of high frequencies of the 0 allele in the west versus high frequencies of the A allele in the east. This pattern is augmented to some extent by high Rh(D) frequencies in the west and high Rh(d) frequencies in the east. The distinctiveness of Longford and Roscommon is associated with high frequencies of B and Rh(D).

Thus, the general picture of the genetic map is a west-east division with two noticeable deviations from this pattern: (1) the divergence of Longford and Roscommon from other mid-land populations, and (2) the extreme divergence of the Aran Islands from the west-coast populations. Alternative explanations are explored in the discussion below.

The first two eigenvectors of the d E matrix based on anthropometric data are shown in figure 4, where both axes have been scaled by the square root of their corresponding eigenvalues. Axis 1 accounts for 44-6070 of the total among-group variation and, excluding the Aran Islands (1) and the Longford-Westmeath (11)

14 l~ 0 0 • 16

5 • 13 • 6 19

0 0 1 o • • • • 2 • 9 1~ • 1 2 4 • 15

0 3

0~

• 7

1

Figure 4. Plot of the first two eigenvectors of the d 2 matrix based on anthropometrics. Each axis is scaled by the square root of the corresponding eigenvatue.

328 J. H. Relethford

subgroups, shows a distinct west-east separation. The west-coast populations (2, 3, 4, 5 and 10) are clustered on the extreme left, while populations on or adjacent to the east coast (12, 15, 16, 18 and 19) are clustered on the right. Midland populations tend to plot near the centre of the picture. The second axis accounts for 32-6°70 of the total variation, and acts to separate the Aran Islands and the Longford-Westmeath subgroups. The third axis (not shown) accounts for only 9-8o70 of the total variation and acts to separate these two subgroups even more. An analysis using only craniofacial variables gave essentially the same results, suggesting that variation in body measurements does not enhance or obscure the basic relationships. Overall, the results from the anthropometric distance analysis supports the idea of a basic west-east separation, along with the distinctiveness of the Aran Islands and certain mid-land populations.

5. Discussion While there are expected differences between the blood-group and

anthropometric analyses, there are several common features which may be examined in light of the population history of Ireland. As indicated above, the separation of populations along a west-east division suggests a correspondence of genetic affinities with geographic distance and/or historic patterns of population movement. Earlier work (Hooton and Dupertuis 1955, Dawson 1964) has shown this west-east pattern on a univariate basis, b u t did not assess alternative hypotheses. Nor did they combine results in a multivariate way. While geographic distance could explain the basic west-east gradient, there is no north-south gradient evident in the present results. Historic population movements can be used to explain the overall trends noted. The invasion model of early Irish occupation (until about 2000 B.P.) states that the original Celtic inhabitants came in four succesive 'waves', whereby each new wave pushed earlier inhabitants further westward (O'RahiUy 1946). Similar arguments have been used to explain the west-east gradient on a univariate basis (Hooton and Dupertuis 1955, Dawson 1964).

Western and eastern regional differences were probably enhanced by additional historic population movements. Some events were probably minimal in effect, such as the Anglo-Norman invasion of 1169 which involved few men (Sunderland et al. 1973). After this time, however, there was an increase in Anglo-Norman invasion and settlement until about 1300, occurring for the most part in eastern Ireland. Subsequent invasions were small in number until the influx of Scottish and English Protestants into Northern Ireland and the eastern portions of the Republic as the result of the Articles of Plantation of James I in 1609 (Hackett, Dawson and Dawson 1956).

The patterns of biological variation shown here do not rule out geographic distance as a factor influencing regional population structure, merely isotropic migration with respect to geographic distance. There are several studies which demonstrate that west-east migration was common during the last century, due mostly to increasing rural-urban migration with Dublin as a prime target of many migrants (Geary and Hughes 1970, Gillman 1970, Kennedy 1973). Such migration seems unidirectional in nature. As such, recent migration might have acted to counter, and not enhance, previous patterns of population movement associated with inhabitation of western regions.

This recent west-east migration might explain the fact that the anthropometric map shows greater distinctiveness of the west-coast populations than the blood-

Genet ic s tructure o f I re land 329

group map. Previous studies have suggested that anthropometr ics are less susceptible to recent events, and tend to show long-term patterns to a greater extent than serological traits (see Relethford and Lees 1982 for a review of these studies). Another possibility is that too few loci are available for the blood-group analyses to show as clear a pattern.

The distinctiveness of the Aran Islands, shown by both blood-group and anthropometric analyses, can be explained in terms of historical events. Original inhabitation of the islands seems to have resulted f rom continued westward movements of the Irish population (Messenger 1969). In the lately sixteenth century the island took on strategic value and English soldiers were frequently garrisoned. The first occurrence was in 1587 when 20 soldiers were placed on the islands at the command of Queen Elizabeth. Following this, there were several times when English soldiers were sent to the islands, often in groups of several hundred, until the late seventeenth century. Many soldiers stayed on the islands to marry local women. Thus, a repeated influx of English men would have had considerable genetic impact, particularly since such events occurred when the local population size was low (Haddon and Browne 1893, Hacket t and Folan 1958).

English affinities of the Aran Islanders has been suggested in several earlier works (Hooton and Dupertuis 1955, Hacket t and Folan 1958) and is confirmed by the present study. Specific affinities are difficult to determine f rom anthropometr ic data. The blood-group data show the Aran Islands are genetically most similar to the east-coast populations of Ireland, as expected f rom a model of English admixture. Further confirmation is provided by an examination of the R matrix based on four populations: the Aran Islands, the Republic of Ireland, Northern Ireland and England. English gene frequencies were taken from Race and Sanger (1968). The first two axes of the resultant eigenvectorial plot is shown in figure 5. The first axis accounts for 93.9°70 of the total variation, and clearly illustrates the relationship of the Aran Islands to England. Taking the Aran Islands as a dihybrid population, a rough estimate of the relative parental contributions can be made. First, the R matrix is converted into a distance matrix following the t ransformat ion suggested by Harpending and Jenkins (1973) (di~ = rii + rjj - 2rij ). Second, the distances between the hybrid and parental populations can be used to estimate admixture. Using the method of Morton, Dick, Allan, Izatt, Hill and Yee (1977), the estimates of admixture are 41°70 Irish and 59% English. The method of Lees and

Aran

Islands

Ireland

Northern

I re I and

• + England

Figure 5. Plot of the first two eigenvectors of the R matrix comparing the Aran Islands to other populations. Each axis is scaled by the square root of the corresponding eigenvalue.

330 J. H. Relethford

Relethford (1978) yield similar estimates of 43% Irish and 57% English. Again, these estimates are crude first-order approximations, but they do suggest an interesting population for further admixture studies.

Both blood-group and anthropometric analyses suggest that genetic distinctiveness of several midland populations. For the blood-group analysis, the counties of Longford (13) and Roscommon (8) are separate from nearby populations. For the anthropometric analysis, the county subgroup of Longford- Westmeath (11) is quite distinct. Hooton and Dupertuis (1955) noted the distinct nature of this subgroup and suggested two explanations: (1) age-related variation in that this subgroup was on average the youngest, and (2) effects of early Viking inhabitation in the area. In their original analysis, Hooton and Dupertuis noted that the young average age of men in Longford-Westmeath could account for some of the group's distinctiveness. In the present study, however, this separation remains even after variables most affected by age are removed. The hypothesis of Viking contact has historical validity. Early Norse invasions of Ireland began in the late ninth century, and the major settlements sprang up in the coastal ports of Wexford, Waterford, Dublin, Cork and Limerick. The only significant inland occupation of the Vikings seems to have occurred around Lough Ree, a lake bordering Longford, Roscommon and Westmeath. This settlement seems to have resulted from expansion of the Limerick Norse up the Shannon River (Hooton and Dupertuis 1955, Jones 1968, Orme 1970). In fact, the Gaelic term 'longphort' (the Gaelic equivalent of Longford) means a Norse fortification (Orme 1970). The relationship of history to naming of the county is not known, however. That other Norse settlements do not show any distinction in blood-group or anthropometric analyses may be due to the fact that these were coastal ports, and as such experienced greater immigration in later times. That is, Norse colonization is a slight factor relative to later historical events. The Norse may have had a greater genetic effect in the midlands.

Even though the blood-group and anthropometric analyses are similar with respect to Longford, there are still several differences which must be considered when evaluating the Viking hypothesis. Roscommon is divergent along with Longford in the blood-group analysis, but the county subgroup containing Roscommon (8) in the anthropometric analysis is not. Likewise, Westmeath (14) is not divergent in the blood-group analysis. The probable cause of these differences lies in the fact that the two analyses use different sampling units. For example, Roscommon is combined with the eastern part of Galway in the anthropometric analysis. When different regions are pooled, the overall relationship matrix may be changed. In any case, the present paper can offer no conclusions, but only suggestions for further studies in the Irish midlands, particularly with a much larger set of variables.

In sum, the present paper shows that the regional population structure of Ireland has been affected by a series of historical population movements. The basic pattern of a west-east gradient suggests a series of genetically different immigrants over a long period of time. Deviations from this pattern are noticed in the Aran Islands and the mid-lands, both of which have historical explanations. A final question to be examined is" how does regional population structure compare with local population structure?

Two previous studies of local population structure allow some comparison of levels. Both of these studies are based on local variation in areas of western Ireland: (1) rural mainland western Ireland during the 1930s (Relethford 1980b, Relethford,

Genetic structure of Ireland 331

Lees and Crawford 1980, 1981), and (2) west-coast islands and isolates in the 1890s (Relethford 1982). In both studies, geographic distance was found to be a major determinant of population structure, with good fits to the expected isolation by distance model. However, the influence of geographic distance was much stronger in the west coast isolates. On the mainland, other factors, such as population size, influenced the degree of migration and genetic similarity. Thus, in the most isolated areas geographic distance remains the single most powerful factor influencing population structure. In a mainland situation geographic distance remains important, but is also supplemented by factors such as the 'pull' of larger populations in accordance with gravity models of migration. Finally, on a regional level, previous historic patterns hold, although continued patterns of unidirectional migration toward Dublin may soon act to counter the basic pattern of variation.

It is clear that much work remains to be done in Irish population structure, both on a local and regional level. The results presented here may be taken less as definite conclusions, and more as suggestions for future research. The uniqueness of Irish history and demography offers a variety of prospects in human population biology.

References DAWSON, G. W. P., 1964, The frequencies of the ABO and Rh(D) blood groups in Ireland from a sample

of 1 in 18 of the population. Annals o f Human Genetics, 28, 49-59. DODINVAL, P. A., 1970, Population structure of A, B, O, AB blood groups in Belgium. Human Heredity,

20, 169-177. FRIEDLAENDER, J. S., 1975, Patterns o f Human Variation (Cambridge, Massachusetts: Harvard

University Press). GEARY, R. C., and HUGHES, J. G., 1970, Internal migration in Ireland. Economic and Social Research

Institute Paper, No. 54, 1-67 GILLMAN, C. J., 1970, County migration: an alternative approach. Economic and Social Research

Institute Paper, No. 54, 68-82. GOWER, J. C., 1966, Some distance properties of latent root and vector methods used in multivariate

analysis. Biometrika, 53, 325-338. HACKETT, W. E. R., and DAWSON, G. W. P., 1958, The distribution of the ABO and Rhesus (D) blood

groups in the Republic of Ireland from a sample of 1 in 37 of the adult population. Irish Journal o f Medical Science, March, 99-109.

HACKETT, W. E. R., and DAWSON, G. W. P., and DAWSON, C. J., 1956, The pattern of the ABO blood group frequencies in Ireland. Heredity, 10, 69-84.

HACKETT, W. E. R., and FOLAN, M. E., 1958, The ABO and Rh blood groups of the Aran Islanders. Irish Journal o f Medical Science, June, 247-261.

HADDON, A. C., and BROWNE, C. R., 1893, The ethnography of the Aran Islands, County Galway. Proceedings o f the Royal Irish Academy, 2, 768-830.

HARPENDING, H., and JENKINS, T., 1973, Genetic distance among Southern African populations. In Methods and Theories o f Anthropological Genetics, edited by M. H. Crawford and P. L. Workman (Albuquerque: University of New Mexico Press), pp. 177-199.

HOOTON, E. A., and DUPERTUIS, C. W., 1955, The PhysicalAnthropology o f Ireland. Peabody Museum Papers (Cambridge, Massachusetts: Peabody Museum), vol. 30, DOS. 1-2.

IMAIZUMI, Y., and MORTON, N. E., 1969, Isolation by distance in Japan and Sweden compared with other countries. Human Heredity, 19, 433-443.

JONES, G., 1968, A History o f the Vikings (New York: Oxford University Press). JORDE, t . B., 1980, The genetic structure of subdivided human populations. In Current Developments in

Anthropological Genetics. Volume 1, Theory and Methods, edited by J.H. Mielke and M.H. Crawford (New York: Plenum Press), vol. 1, pp. 135-208.

JORDE, L. B., ERIKSSON, A. W., MORGAN, K., and WORKMAN, P. L., 1982, The genetic structure of Iceland. Human Heredity, 32, 1-7.

KENNEDY, R. E., JR., 1973, The Irish: Emigration, Marriage, and Fertility (Berkeley, California: University of California Press).

KOPE¢, A. C., 1970, The Distribution o f Blood Groups in the United Kingdom (London: Oxford University Press).

LEES, F. C., and RELETHFORD, J. H., 1978, Admixture estimation using skin reflectance data. American Journal o f Physical Anthropology, 49, 505-510.

332 J. H. Relethford

LEES, F. C., and RELETHFORD, J. H., 1982, Population structure and anthropometric variation in Ireland during the 1930s. In Current Developments in Anthropological Genetics, Volume 2: Ecology and Population Structure, edited by M. H. Crawford and J. H. Mielke (New York: Plenum Press), pp. 385--428.

MESSENGER, J. C., 1969, Inis Beag: Isle of Ireland (New York: H01t, Rinehart and Winston). MORTON, N. E., DICK, H. M., ALLAN, N. C., IZATT, M. M., HILL, R., and YEE, S., 1977, Bioassay of

kinship in northwestern Europe. Annals of Human Genetics, 41, 249-255. MORTON, N. E., KLEIN, D. HUSSELS, I. E., DODINVAL, P., TODOROV, A., LEW, R., and YEE, S., 1973,

Genetic structure of Switzerland. American Journal of Human Genetics, 25, 347-361. MORTON, N. E., YEE, S., HARRIS, D. E., and LEw, R., 1971, Bioassay of kinship. Theoretical Population

Biology, 2, 507-524. O'RAHILLY, T. F., 1946, Early Irish History and Mythology (Dublin: Institute for Advanced Studies). ORME, A. R., 1970, Ireland (Chicago: Aldine Publishing Co.). RACE, R. R., and SANGER, R., 1968, Blood Groups in Man (Philadelphia: F. A. Davis), 5th edition. RELETHFORD, J. H., 1980 a, Bioassay of kinship from continuous traits. Human Biology, 52, 689-700. RELETHFORD, J. H., 1980 b, Population Structure and Anthropometric Variation in Rural Western

Ireland. Ph.D. thesis, Department of Anthropology, State University of New York at Albany. RELETHFORD, J. H., 1982, Isonymy and population structure of Irish isolates during the 1890s. Journal of

Biosocial Science, 14, 241-247. RELETHEORD, J. H., and LEES, E C., 1981, The effects of aging and secular trend on adult stature in rural

western Ireland. American Journal of Physical Anthropology, 55, 81-88. RELETHFORD, J. H., and LEES, F. C., 1982, The use of quantitative traits in the study of human

population structure. Yearbook of Physical Anthropology, 25, 113-132. RELETHFORD, J. H., and LEES, F. C., 1983, Correlation analysis of distance measures in Ireland based on

geography, anthropometry, and isonymy. Human Biology (in the press). RELETHFORD, J. H., LEES, F. C., and CRAWFORD, M. H., 1980, Population structure and anthropometric

variation in rural western Ireland: migration and biological differentiation. Annals of Human Biology, 7, 411-428.

RELETHFORD, J. H., LEES, F. C., and CRAWFORD, M. H., 1981, Population structure and anthropometric variation in rural western Ireland: isolation by distance and analysis of the residuals. American Journal of Physical Anthropology, 55, 233-245.

ROBERTS, D. F., JORDE, L. B., and MITCHELL, R. J., 1981, Genetic structure in Cumbria. Journal of Biosocial Science, 13, 317-336.

SUNDERLAND, E., TILLS, n . , BOULOUX, C., and DOYL, J., 1973, Genetic studies in Ireland. In Genetic Variation in Britain, edited by D. F. Roberts and E. Sunderland (London: Taylor & Francis), pp. 141-159

TILLS, D., 1977 a, Red cell and serum proteins and enzymes of the Irish. Annals of Human Biology, 4, 35-42.

TILLS, D., 1977 b, The use of the FST statistic of Wright for estimating the effects of genetic drift, selection and migration in populations, with special reference to Ireland. Human Heredity, 27, 153-159

TILLS, D., TEESDALE, P., and MOURANT, A. E., 1977, Blood groups of the Irish. Annals of Human Biology, 4, 23-24.

WALTER, H., and PALSSON, J., 1973, The incidence of some genetic markers in Ireland. In Genetic Variation in Britain, edited by D. F. Roberts and E. Sunderland (London: Taylor & Francis), pp. 161-170.

WORKMAN, P. t . , HARPENDING, H., LALOUEL, J. M., LYNCH, C., NISWANDER, J. D., and SINGLETON, R., 1973, Population studies on southwestern Indian tribes. VI. Papago population structure: A comparison of genetic and migration analyses. In Genetic Structure of Populations, edited by N.E. Morton (Honolulu: University Press of Hawaii), pp. 166-i94.

WORKMAN, P. L., MIELKE, J. H., and NEVANLINNA, H. R., 1976, The genetic structure of Finland. American Journal of Physical Anthropology, 44, 341-368.

ZANARDI, P., DELL'ACQUA, G., MENINI, C., and BARRAI, I., 1977, Population genetics in the province of Ferrara. I. Genetic distances and geographic distances. American Journal of Human Genetics, 29, 169-177.

Address correspondence to: Dr John H. Relethford, Department of Anthropology, State University of New York at Oneonta, Oneonta, NY 13820, USA.

Zusammenfassang. Bev61kerungsstruktur und Geschichte kOnnen auf einem lokalen oder regionalen Niveau untersucht werden. Diese Arbeit untersucht die regionale BevOlkerungsstruktur der Republik Irland und von Nordirland in bezug auf BevOlkerungsgeschichte und demographische Prozesse. Blutgruppen- und anthropometrische Daten aus der Literatur wurden analysiert. Die Blutgruppendaten bestehen aus Genfrequenzen ftir ABOund Rhesus ftir 32 Grafschaften und die Aran-lnseln. Die anthropo-

Genetic structure o f Ireland 3 3 3

metrischen Daten bestehen aus Gruppenstat ist iken ft~r fi3nfzehn Variable aus 19 Regionen. A u s m a g und Struktur der Bev01kerungsdifferenzierung wird mithilfe neuer Methoden der BevNkerungss t rukturanalyse erfal3t. Sowohl Blutgruppen als auch anthropometr ische Analysen zeigen eine West-Ost-Einteilung der Bev01kerung, was der bekanuten Einwanderungsgeschichte von Irland entspricht, in der aufeinanderfol- gende Wellen yon Einwanderern frtihere Bev01kerungen weiter nach Westen drtickten. In beiden Analysen gab es zwei Abweichungen von diesem Grundmuster : die Aran-Inseln und das Mittelland. In beiden Fallen werden alternative geschichtliche Erkl~.rungen geprOft. Das genetische Verh~iltnis der Aran-Inseln zum Rest von Irland und England scheint mit der englischen Beimischung zusammenzuh~ingen, die der Stationierung yon Soldaten vor einigen Jahrhunder ten folgte. Die genetische Posit ion des Mittellandes ist komplexer, sie legt aber die Wirkung einer fraheren WikingerbevOlkerung nahe. Diese Ergebnisse sind mit Lokaluntersuchungen und weniger mit Regionaluntersuchungen zur Bev61kerungsstruktur von Irland verknt~pft.

R6sum6. La structure et l 'histoire des populat ions peuvent &re 6tudi6es au plan local ou au plan r6gional. Cet article examine la structure populationelle r6gionale de la R6publique de l ' Ir lande et de l ' I r lande du Nord par rapport h l 'histoire des populations et les processus d6mographiques. Sont analys6es les donn6es de la litt6rature sur les groupes sanguins et l 'anthropom6tr ie . Les donn6es sur les groupes sanguins consistent en fr6quences g6niques ABO et Rh6sus pour 32 comt6s et les lles Aran. Les donn6es anthropom6triques consistent en statistiques condens6es pour quinze variables r6colt6es en 19 r6gions. Le degr6 et la configuration de la diff6renciation des populat ions sont 6tablis ~ l 'aide de nouvelles m6thodes d 'analyse de la structure des populations. Aussi bien l 'analyse des groupes sanguins que celle de l 'anthropom6trie montrent une division est-ouest des populat ions correspondant & l 'histoire connue du peuplement de l 'Irlande, off des vagues successives d ' immigran ts ont refoul6 plus ~ l 'ouest les populat ions anterieures. Dans chacune des deux analyses il y avait deux 6carts ~ cette configuration de base: les Iles Aran et les midlands. Dans les deux cas, des explications historiques alternatives sont examin6es. La relation g6n6tique des Iles Aran au reste de l ' Ir lande et ~t l 'Angleterre semble dfie ~ une admixion anglaise cons6cutive au casernement de soldats il y a plusieurs si~cles. La position g6n6tique des midlands est plus complexe, mais sugg~re les effets d ' u n peuplement Viking ancien. Ces r6sultats sont mis en relation avec des 6tudes au plan local p lut f t que r6gional de la structure des populations de l ' lr lande.