The Origins of the British: The New Prehistory of Britain (76 page)

30
. Procopius,
History of the Wars
8.20.6–10.

31
. Cruciform brooches are also absent from Lower Saxony: see e.g. Scarre (1995), p. 180, map 1.

1
. i.e. from the northern Germanic region, including Denmark and farther north.

2
. e.g.
The Anglo-Saxon Chronicle
, which was originally compiled on the orders of King Ælfred the Great, c.
AD
890, and continued into the Middle Ages.
Part 1
(entry for the year 495) contains a list of the Saxon Kings of England, mostly lifted from Bede. Translation available at the Online Medieval and Classical Library:
http://omacl.org/Anglo/part1.html
>.

3
. There is also a genealogy mentioned in
Beowulf
. The date of
Beowulf
has yet to be determined.

4
.
The Anglo-Saxon Chronicle
, entry for the year
AD
449.

5
. Heyerdahl and Lillieström (2001).

6
. The issue of ‘Old Saxony’ during Roman times and its likely location at the base of the Cimbrian Peninsula and on the east bank of the Elbe has been addressed by several authors, who independently came to the same conclusion. There is textual evidence for this, both at the time of Augustus’ death in
AD
14 (Treharne and Fullard 1976, plates 10 and 15, map 2a), and in Ptolemy’s
Geographia
(c.
AD
150). See also Chadwick (1907), figure on p. 195; Forster et al. (2006), figure 11.2; and Forster (1995).

7
. Chadwick (1907), p. 112; Treharne and Fullard (1976), Medieval section, p. 7.

8
. Saxo Grammaticus,
Gesta Danorum
, late twelfth to early thirteenth century.

9
. Saxo Grammaticus, Book 1.

10
. Anderson (1736).

11
. This formula (declaration) of renunciation is dated
AD
743 (Vulpius 1826).

12
. Wrenn and Bolton (1988), pp. 25–33.

13
. Wrenn and Bolton (1988), p. 56.

14
. Wrenn and Bolton (1988), p. 38.

15
. The Winchester MS is also known as the
The Parker Chronicle
(Corpus Christi College, Cambridge, MS. 173).

16
. Bede,
Ecclesiastical History
2.15.

17
. Clarke (1960), p. 138ff.

18
. Newton (1992); see also Newton (1993).

19
. Gildas,
De excidio Britanniae
23.

20
. Newton (2003).

21
. Carver (1998), pp. 32, 36, 43, 47, 56, 128.

22
. Carver (1998), p. 36.

23
. Carver (1998), p. 38.

24
. Hines (1992).

25
. Myhre (1992).

26
. Hills (1998), Weber (1998).

27
. Dumville (1989).

1
. Traditional archaeological view of replacement: e.g. Myres (1969, 1986), Leeds (1912). Still in favour of significant demic impact: e.g. Härke (2002). Geneticists in favour of Anglo-Saxon ‘replacement’: e.g. Weale et al. (2002).

2
. Higham (1992), Hamerow (1997); see also earlier sceptics: Arnold (1984) and Hodges (1989).

3
. Crawford (1997).

4
. See e.g. Härke (2002).

5
. Hamerow (1997).

6
. Various figures have been estimated, but no higher than 5%: e.g. 1% (Härke 2002) and 3–5% (Bragg 2003, p. 42).

7
. Diamond (1997).

8
. Pryor (2004), pp. 143–4.

9
. Hamerow (1997), p. 37.

10
. Hamerow (1997), p. 40.

11
. Myres (1986), pp. 176–8.

12
. Myres (1986), p. 14.

13
. Cavalli-Sforza et al. (1994).

14
. Watkin and Mourant (1952).

15
. Watkin (1966).

16
. Viereck (1998).

17
. Kope (1970). Note that similar arguments have been put forward for Rhesus negative rates, which are high in the Basque Country. To keep things simple in this discussion of method, I do not refer to these.

18
. Viereck (1998).

19
. See e.g. Daniels (2002).

20
. Weale et al. (2002).

21
. For an accessible edition see Hinde (2004).

22
. In more detail: ‘Firstly, little genetic differentiation exists among the Central English towns … Secondly, in contrast to the Central English towns, the two North Welsh towns show highly significant differences, both from each other and from the five Central English towns … Thirdly, no significant differences in haplotype frequencies exist between Friesland and any of the Central English towns. Comparisons between Norway and the Central English towns, on the other hand, are all significant, apart from Bourne … Similar results were obtained using
F
_ST
values based on haplogroup frequencies, but tests on
F
_ST
values based on haplotype frequencies were not significant because of the large number of singletons at this level … Taken together, these results suggest considerable male-line commonality between Central England and Friesland’ (Weale et al. 2002).

23
. In spite of the claims by Weale et al. (2002) that they were using 150 highly differentiated gene types to make comparisons, the ‘statistical significance’ tests required to make the inferences of ‘extremely high affinity’ rested on calculations based on only seven haplogroups rather than many haplotypes.

24
. Thomas et al. (2006).

25
. Capelli et al. (2003).

26
. Wilson et al. (2001).

27
. The stated reason for amalgamating these samples was that there were ‘no significant differences between them’ (Capelli et al. 2003). However, the small
number of samples (23) from the north-west Germany source (Lower Elbe river) may have been more pertinent. There were 100 samples from Denmark and 62 from Schleswig-Holstein.

28
. To improve on the haplogroup number and resolution, Capelli et al. (2003) added several more UEP (unique event polymorphisms – see below) markers and also created three STR-defined clusters of one-step relatives around the three of the more common haplotypes: the Atlantic Modal Haplotype and two others (Ht. 157, 393 and 96 in this study). In this context, haplogroup is defined by unambiguous markers in the Y-chromosome, referred to by Capelli et al. as UEPs and more generally in the literature as ‘bi-allelic’ (i.e. dichotomous) markers: see Underhill et al. (2000) and Semino et al. (2000). The three extra clusters are defined by STR (single tandem repeat polymorphisms), which are less stable and not ‘unambiguous’. They correspond, respectively, to the clusters R1b-10, I1a-4 and R1a1-3, within haplogroups R1b, I and R1a1, which I have used in this book. The difference is that while Capelli et al. have chosen only three clusters, thus adding to their 11 UEP-defined haplogroups to make 14 groups, I have systematically broken up all the large haplogroups into clusters in this way (see Appendix C), thus adding 30 clusters to 11 haplogroups and making 41 groups.

29
. Plot based on Principal Components Analysis using their 14 groups and displayed their results in the form of a two-dimensional genetic distance map using first two Principal Components (Capelli et al. 2003).

30
. They also note that ‘All continental populations, however, show significant differences from the indigenous group (
p
0.01)’ (Capelli et al. 2003).

31
. In a separate Principal Components plot using simulated populations based on various theoretical admixtures of Norwegian, NGD and ‘indigenous British gene groups’, Capelli et al. (2003) show a very similar outcome which illustrates where each level of admixture would place the simulated populations. By analogy, this essentially places the majority of the sampled British market towns on the ‘indigenous’ side of the plot (i.e. less than 50% intrusion).

32
. Capelli et al. (2003) do address this point: ‘With regard to source populations, we note that Weale et al. … recently used Friesland as an Anglo-Saxon representative source population and suggested a substantial replacement of pre-Anglo-Saxon paternal lineages in central England. We therefore compared Frisians to our North German/Danish sample and found that the two sets are not significantly different from each other (
p
= 0.3, data not shown). When
included in the PC analysis, the Frisians were more “Continental” than any of the British samples, although they were somewhat closer to the British ones than the North German/Denmark sample.’

I repeated the analysis and was able to confirm that the Frisians were indeed intermediate and closer to the British than to the NGD sample. However, the lack of significant difference between Frisians and the NGD sample in haplogroup frequency is to be expected, given the overall genetic similarity and shallow regional geographical gradients between
all
the north-west European populations at this level of haplogroup resolution. For ‘shallow geographic gradients’ at the haplotype level, see figure 4a–c in Roewer et al. (2005), who look at STR variation geographically, but do not use it to resolve the relevant haplogroups.

33
. Rosser et al. (2000). While using slightly different UEP markers and 12 haplogroups, and defining slightly fewer haplogroups than in the Weale or Capelli studies, all the important divisions are present in the Rosser database and additionally they can be closely matched from the Weale and Capelli English databases (Hg1 = R1b and PxR1a1; Hg2 = I and FxJK; Hg3 = R1a1; Hgs 4,8,12 = E3; Hg9 = J; Hg16 = N3; Hg26 = KxPN3); in addition they include five British locations. See also Roewer et al. (2005); they include only two British sample points, have a different selection of UEPs and STRs and avoid using the phylogenetic information, but their analysis still, like Rosser’s, shows three poles of variation (represented as three ‘dimensions’).

34
. Amos et al. (2006) state that: ‘Two previous studies have used large datasets to explore the male genetic ancestry of Britain [Weale et al. (2002), Capelli et al. (2003)] … Both studies reveal interesting patterns, but both are limited by a combination of using a small number of extrinsic continental outgroups (three), uncertainties about the timing of events and the perennial question of who is mixing with whom. To examine the effectiveness of the mixing profile approach to elucidate complicated population histories, we therefore reanalysed the larger dataset of Capelli et al., augmenting it with further data from other continental populations.’