Figure 3: Simplified schematic plan to show sampled timbers
(taken at tiebeam level with rafters superimposed)
Site
North
Truss 1
Frontage and
driveway
6
5
4
Bay1
Truss 2
8
3
7
Bay 2
2
1
Truss 3
Figure 4: Bar diagram of the samples in site chronology CYNASQ01
Offset
20
25
31
25
48
42
00
no h/s
no h/s
A05
A03
h/s
h/s
A02
A06
h/s
A07
20C sap
20C sap
H03
A08
00
20
40
60
80
100
Total
rings
Relative
heartwood/sapwood
boundary position
62
60
64
74
57
78
120
95
99
105
100
100
120 years relative
White bars = heartwood rings, shaded area = sapwood rings
h/s = the last ring on the sample is the heartwood/sapwood boundary, only the sapwood is missing
C = complete sapwood is retained on the sample, the last measured ring date is the felling date of the timber
Table 1: Details of samples from a cruck building at The Celyn, Lower Maescoed, Herefordshire
Sample
number
CYN-A01
CYN-A02
CYN-A03
CYN-A04
CYN-A05
CYN-A06
CYN-A07
CYN-A08
CYN-H03
Sample location
North common rafter 9, bay 2
North common rafter 7, bay 2
North common rafter 2, bay 2
North common rafter 6, bay 1
North common rafter 4, bay 1
North common rafter 1, bay 1
North common rafter 5, bay 2
North cruck blade truss 2
South cruck blade truss 2
Total
rings
*Sapwood
rings
First measured
ring date
Last heartwood
ring date
Last measured
ring date
nm
64
60
54
62
74
57
120
78
h/s
h/s
no h/s
no h/s
no h/s
h/s
h/s
20C
20C
-----1378
1372
-----1367
1372
1395
1347
1389
-----1441
---------------1445
1451
1446
1446
-----1441
1431
-----1428
1445
1451
1466
1466
*h/s = the last ring on the sample is at the heartwood/sapwood boundary
no h/s = the sample does not have the heartwood/sapwood boundary
nm = sample not measured
C = complete sapwood is retained on the sample, the last measured ring date is the felling date of the timber
T-value off-set matrix showing the cross matches
between individual samples
2
2
3
5
6
7
8
10.3
2.4
5.5
1.1
4.1
3
6
5
-10
5
9.6
6.8
0.9
4.7
8.4
1.3
4.1
6
6
0
-5
2.1
4.6
7
-3
16
-4
1
8
31
25
20
11
48
3.9
(Off-sets above diagonal, t-values below diagonal)
Table 2: Results of the cross-matching of site chronology CYNASQ01 and relevant reference
chronologies when the first ring date is 1347 and the last ring date is 1466
Reference chronology
Span of
chronology
t-value
Tusmore Park, Oxon
Kingswood Abbey Gatehouse, Glos
MC10---H
Gotham Manor, Notts
Southern England
Wales and West Midlands
Mercer’s Hall. Gloucester
East Midlands
England
Sinai Farm, Burton on Trent, Staffs
1359 - 1545
1307 - 1428
1386 - 1585
1410 - 1474
1083 - 1981
1341 - 1636
1289 - 1541
882 - 1981
401 - 1981
1336 - 1499
6.9
6.2
6.2
6.2
6.1
5.7
5.7
5.5
5.3
5.2
( Howard et al 1992 )
( Arnold et al 2003 )
( Fletcher, 1978 unpubl )
( Howard et al 1991 unpubl )
( Bridge 1988 )
( Siebenlist-Kerner 1978 )
( Howard et al 1996 )
( Laxton and Litton 1988 )
( Baillie and Pilcher 1982 unpubl )
( Howard 2004 unpubl )
TREE-RING ANALYSIS OF TIMBERS FROM
A CRUCK BUILDING AT THE CELYN,
LOWE MAESCOED,
LONGTOWN,
HEREFORD
A J ARNOLD
R E HOWARD
TREE-RING ANALYSIS OF TIMBERS FROM
A CRUCK BUILDING AT THE CELYN,
LOWE MAESCOED,
LONGTOWN,
HEREFORD
A J ARNOLD
R E HOWARD
Summary
Core samples were obtained from seven different rafters and one blade within a
cruck building at The Celyn, Lower Maescoed. Analysis of these cores, in
conjunction with data obtained during a previous programme of tree-ring dating
at The Celyn produced a single site chronology, CYNASQ01, comprising seven
samples and having an overall length of 120 rings. These 120 rings were
satisfactorily dated as spanning the years 1347 to 1466.
Interpretation of the sapwood on the dated samples would indicate that all the
cores represent timbers felled between late 1466 and early 1467, and that the
roof, which in any case appears to be structurally of one build and integral with
the body of the house, is the same date as the main building timbers.
Introduction
Set impressively against the dramatic backdrop of the Black Mountains in
Herefordshire, The Celyn presents a now floored, but, as evidenced by the smoke
blackening to the rafters, once open two-bay, stone-built cruck building, Figure 1. Three
substantial cruck trusses remain, that to the east gable, truss 1, having a single collar,
while truss 2 has two collars (Fig 2). The collar of truss 3, at the west end, is tennoned
into the blades. These cruck trusses carry double purlins, which in turn support the
common rafter roof.
Sampling
There appears to have been considerable interest and debate amongst interested
parties as to the date of the cruck phase of the house, and as to how much of the
original building now remains. This discussion initially focused on the cruck trusses,
which have been dated previously by tree-ring analysis as being felled in 1466/7 (Miles
and Worthington 2003). Of late there has been further interest in the roof which, it has
been suggested, might not be original but a later replacement. Thus, in order to bring
certainty and reliability to the discussion, tree-ring analysis of the rafters in particular
was commissioned by the owner, Mr Thomas Hill.
An assessment of the roof timbers, the common rafters between the cruck trusses,
shows them to be moderately large, regularly cut, and quite heavily trimmed, flat-faced
beams, about 15 - 17 cms by 8 to 10 cms. The great majority of these timbers are
derived from relatively young, fast grown trees. As such they generally have low
numbers of rings and most of them are unsuitable for tree-ring dating. It was thus purely
fortuitous that sufficient samples could be obtained.
From the material available a total of eight core samples was obtained. Seven of these
cores were obtained from different common rafters while a further single sample was
obtained from one of the cruck blades. This eighth core was taken to help provided
sufficient site data to be able to compile a reliable 'site master sequence'. Each sample
was given the code CYN-A (for Celyn, site 'A'), and numbered 01 - 08. Details of the
samples are recorded in Table 1, with the locations of the samples also being recorded
on a simple sketch plan made at the time of sampling, reproduced here as Figure 3. In
this report the trusses have been numbered from site east (the gable end) to west, with
timbers being identified on a north - south basis as appropriate.
Tree-ring dating
Tree-ring dating relies on a few simple, but quite fundamental, principles. Firstly, as is
commonly known, trees (particularly oak trees, the most frequently used building timber
in England) grow by adding one, and only one, growth-ring to their circumference each,
and every, year. Each new annual growth-ring is added to the outside of the previous
year’s growth just below the bark. The width of this annual growth-ring is largely, though
not exclusively, determined by the weather conditions during the growth period (roughly
March – September). In general, good conditions produce wider rings and poor
conditions produce narrower rings. Thus, over the lifetime of a tree, the annual growth-
rings display a climatically influenced pattern. Furthermore, and importantly, all trees
growing in the same area at the same time will be influenced by the same growing
conditions and the annual growth-rings of all of them will respond in a similar, though
not identical, way.
Secondly, because the weather over any number of consecutive years is unique, so too
is the growth-ring pattern of the tree. The pattern of a short period of growth, 20, 30 or
even 40 consecutive years, might conceivably be repeated two or even three times in
the last one thousand years. A short pattern might also be repeated at different time
periods in different parts of the country because of differences in regional microclimates. It is less likely, however, that such problems would occur with the pattern of a
longer period of growth, that is, anything in excess of the statistically reliable minimum
of 54 years. In essence, a short period of growth, anything less than 54 rings, is not
reliable, and the longer the period of time under comparison the better.
The third principal of tree-ring dating is that, until the early- to mid-nineteenth century,
builders of timber-framed houses usually obtained all the wood needed for a given
structure by felling the necessary trees in a single operation from one patch of
woodland, or from closely adjacent woods. Furthermore, and contrary to popular belief,
the timber was used "green" and without seasoning, and there was very little long-term
storage as in timber-yards of today. This fact has been well established from a number
of studies where tree-ring dating has been undertaken in conjunction with documentary
studies. Thus, establishing the felling date for a group of timbers gives a very precise
indication of the date of their use in a building.
Tree-ring dating relies on obtaining the growth pattern of trees from sample timbers of
unknown date by measuring the width of the annual growth-rings. This is done to a
tolerance of 1/100 of a millimetre. The growth patterns of these samples of unknown
date are then compared with a series of reference patterns or chronologies, the date of
each ring of which is known. When the growth-ring sequence of a sample 'crossmatches' repeatedly at the same date span against a series of different relevant
reference chronologies the sample can be said to be dated. The degree of crossmatching, that is the measure of similarity between sample and reference, is denoted
by a 't-value'; the higher the value the greater the similarity. The greater the similarity
the greater is the probability that the patterns of samples and references have been
produced by growing under the same conditions at the same time. The statistically
accepted fully reliable minimum t-value is 3.5.
However, rather than attempt to date each sample individually it is usual to first
compare all the samples from a single building, or phase of a building, with one
another, and attempt to cross-match each one with all the others from the same phase
or building. When samples from the same phase do cross-match with each other they
are combined at their matching positions to form what is known as a 'site chronology'.
As with any set of data, this has the effect of reducing the anomalies of any one
individual (brought about in the case of tree-rings by some non-climatic influence) and
enhances the overall climatic signal. As stated above, it is the climate that gives the
growth pattern its distinctive pattern. The greater the number of samples in a site
chronology the greater is the climatic signal of the group and the weaker is the nonclimatic input of any one individual.
Furthermore, combining samples in this way to make a site chronology usually has the
effect of increasing the time-span that is under comparison. As also mentioned above,
the longer the period of growth under consideration, the greater the certainty of the
cross-match. Any site chronology with less than about 55 rings is generally too short for
satisfactory analysis.
Analysis
Each of the eight samples obtained from The Celyn was prepared by sanding and
polishing. It was seen at this point that one sample, CYN-A01, had too few rings (less
than 54) for reliable dating and it was rejected from this analysis. The annual growthrings of the remaining seven samples were, however, measured. These data were then
compared with each other and at a relatively high minimum value of t=6.0 six of the
seven samples cross-matched with each other at the relative positions shown in the bar
diagram Figure 4. The six cross-matching samples were combined at these relative
positions to form site sequence CYNASQ01, with a very satisfactory combined overall
length of 120 rings.
Site chronology CYNASQ01 was then reliably dated by comparison with a large
number of relevant reference chronologies for oak as spanning the years 1347 to 1466.
The evidence for this dating is given in the t-values of Table 2 where a selection of
cross-matches is shown.
Interpretation
Analysis by dendrochronology of a small number of samples from The Celyn has
produced a single site chronology, CYNASQ01, comprising six samples, its 120 rings
dated as spanning the years 1347 to 1466. One sample in this site chronology, CYNA08, from one of the cruck blades, retains complete sapwood, denoted by ’C’ in Table 1
and the bar diagram. This means that it has the last ring produced by the tree
represented before it was felled. This last complete sapwood ring, and thus the felling
of the timber, is dated to 1466. However, given the amount of cell growth present for
that year it is possible to say that the tree was probably felled after the summer of that
year and before the spring of 1467
Crucially for the interpretation of the felling date of the other roof timbers, the relative
positions of the heartwood/sapwood boundaries on the other three dated samples,
where it exists, (denoted by 'h/s' in Table 1 and the bar diagram), is consistent with
these timbers also being felled in 1466/7. This boundary varies by only 10 years from
relative position 95 (1441), on sample CYN-A02, to relative position 105 (1451), on
sample CYN-A07. The consistency of the heartwood/sapwood boundary seen in all the
samples is indicative of timbers having a single phase of felling.
Conclusion
Analysis by tree-ring dating of a number of samples from the roof timbers of the cruck
building at The Celyn and one from a cruck blade has shown that they all represent
trees felled at the same time. Given that one sample, that from the cruck blade, has
complete sapwood and has a felling date of 1466, it can be reliably demonstrated that
the roof timbers were felled at this time too. Such a felling date is highly consistent with
that obtained through previous tree-ring analysis of other samples from the crucks. It
would thus appear that not only have the cruck trusses themselves survived for just
almost 540 years, but so too has the roof which, as is intimated by the structural
evidence and smoke blackening, is integral to, and of the same phase as, the main
building timbers.
Judging by the t-values of the cross-matching between some samples, it is possible
that they represent two timbers derived from the same tree. As might be seen in the tvalue/off-set matrix below, this may be the case, for example, with the timbers
represented by samples CYN-A02 and A03. Other samples, ie CYN-A05 and A06, may
represent other trees growing close by in the same patch of woodland. It would thus
appear that all the timbers used come from trees once growing in the same stand or
copse.
Bibliography
Arnold, A, J, Howard, R E, and Litton C D, 2003 Tree-ring analysis of timbers from
Kingswood Abbey Gatehouse, Kingswood, Gloucestershire, Centre for Archaeol Rep,
21/2003
Baillie, M G L, and Pilcher, J R, 1982 unpubl A master tree-ring chronology for England,
unpubl computer file MGB-EOI, Queens Univ, Belfast
Bridge, M, 1988 The Dendrochronological Dating of Buildings in Southern England,
Medieval Archaeol, 32, 166 – 74
Fletcher, J, 1978 unpubl computer file MC10---H
Howard, R E, Laxton, R R, Litton, C D, and Simpson, W G, 1991 unpubl - Nottingham
University Tree-Ring Dating Laboratory, unpubl computer file GOTMSQ06
Howard, R E, Laxton, R R, Litton, C D, and Simpson, W G, 1992 List 44 no 16 Nottingham University Tree-Ring Dating Laboratory: results, Vernacular Architect, 23,
51 – 6
Howard, R E, Laxton, R R, and Litton, C D, 1996 Tree-ring analysis of timbers from
Mercer’s Hall, Mercer’s Lane, Gloucester, Anc Mon Lab Rep, 13/1996
Howard, R E, 2004 unpubl, Nottingham University Tree-ring Dating Laboratory,
unpublished computer file SINBSQ01 for Sinai Farm, Burton on Trent, Staffs
Laxton, R R, and Litton, C D, 1988 An East Midlands master tree-ring chronology and
its use for dating vernacular buildings, University of Nottingham, Dept of Classical and
Archaeol Studies, Monograph Series, III
Miles, D W H, and Worthington, M J, 2002/3 The tree-ring dating of the cruck house
and barn at The Celyn, Longtown, Herefordshire – Oxford Dendrochronology
Laboratory interim report
Siebenlist-Kerner, V, 1978 Chronology, 1341-1636, for hillside oaks from Western
England and Wales, in Dendrochronology in Europe (ed J M Fletcher), BAR Int Ser, 51,
295 – 301
Figure 1: The cruck building at The Celyn
(viewed from the front or ‘site north’)
Figure 2: Truss 2 showing the west face