e-PS, 2008, 5, 29-35

ISSN: 1581-9280 web edition

e-PRESERVATIONScience
ISSN: 1854-3928 print edition

www.Morana-rtd.com published by M O R A N A RTD d.o.o.

© by M O R A N A RTD d.o.o.

A RE-EVALUATION OF THE USE

OF MAXIMUM MOISTURE CONTENT DATA
FOR ASSESSING THE CONDITION
FULL PAPER OF WATERLOGGED ARCHAEOLOGICAL WOOD

Glenn McConnachie *1 , Rod Eaton 2 , Mark Jones 3

1. Conservation Manager, The Mary Maximum moisture content (U max) “profiles” were produced
Rose Trust, College Road, H M Naval
throughout cut cross-sections of waterlogged archaeologi-
Base, Portsmouth, PO1 3LX, U.K.
2. Deputy Head of Department (retired), cal oak, poplar and pine timbers, and the distribution of
School of Biological Sciences, University values compared to visible degradation patterns. Attempts
of Portsmouth, King Henry Building,
were then made to classify the degree of degradation in
King Henry I St, Portsmouth, PO1 2DY,
U.K. each timber. Where appropriate, comparisons were made
3. Head of Collections, The Mary Rose with the established classification scheme of de Jong
Trust, College Road, H M Naval Base,
(1977), for waterlogged archaeological oak. Numerical clas-
Portsmouth, PO1 3LX, U.K.
sification schemes were investigated after statistically
corresponding author:
[email protected]
analysing U max data ranges for each of the three species.

Moisture content profiles from all timber sections reflec-

ted visible degradation patterns, supporting the use of Umax
data as an indicator of preservation state. Incorporating
numerical data in to a classification system presented
potential problems, as large ranges of values were common
within individual timbers. Numerical schemes were sugge-
sted however, by incorporating two ranges of Umax values,
which separate well preser ved and degraded regions
respectively. U max ranges were unique to the individual spe-
cies studied and a generalised scheme was not possible. A
generic classification system based on visual appearance
or physical examination is proposed, by modifying de Jong’s
scheme for European oak, which compares relative propor-
tions of well preser ved versus degraded material.
Collectively, results provide reference data for assessing
similar timbers, where large destructive samples cannot be
taken.

1 Introduction

The design of an appropriate conservation treatment for a waterlog-

ged archaeological wooden object will be dictated largely by its state
received: 09.04.2008 of preservation. An accurate condition assessment is a crucial first
accepted: 06.05..2008
step for conservators, who may employ a number of analytical techni-
key words: ques. Chemical analyses, 1-3 examination of wood microstructure 4,5
moisture content, waterlogged wood,
condition assessment
and measurement of physical and mechanical properties 3,6-10 have all

29
www.e-PRESERVATIONScience.org

CLASS III CLASS II CLASS I

Figure 1: De Jong’s classification for the degree of degradation of European oak. Adapted from ref. 18. CLASS III - “Wood containing <185%
water”. Sound core beneath a thin deteriorated layer. CLASS II – “Wood containing 185 - 400% water”. Comparatively small core present.
CLASS I – “Wood containing >400% water”. Highly degraded wood predominates.

been shown to be useful indicators of the preser- of U max values are also examined, in an attempt to
vation state of waterlogged wooden objects. provide a numerical classification scheme that
relates to a visual or physical assessment of
One simple and routinely employed technique is degradation patterns. Similar data is examined for
the determination of moisture content. Values rise a selection of poplar and pine timbers from the
in line with increases in the porosity of the wood Mary Rose , to assess how degradation patterns,
cell wall, resulting from microbiological degrada- U max ranges and potential classification schemes
tion 11-13 and to a lesser extent from physicochemi- might vary for timbers of other species. The aim is
cal processes 14-16 before and / or during burial. to produce reference data describing typical
Hoffmann 1 stressed the importance of using maxi- degradation patterns and U max ranges for the three
mum, rather than actual moisture content meas- species, which can then be used to aid the asses-
urements. By initially placing submerged samples sment of similar timbers, where destructive samp-
cyclically under partial vacuum, any trapped air is ling is not possible.
expelled. This ensures a maximum moisture con-
tent reading is obtained, that reflects the full
extent of the voids that have resulted from degra-
dation.

Waterlogged archaeological timbers often display

visibly distinct regions of differently degraded Visual appearance of tim-
material in cross-section. In decayed, high moistu- Timber description Ref. Code ber cross-section
re content areas, wood is physically softer and Very thin decayed outer
appears darker, with less distinct surface details. layer and large sound
Oak “Deck Beam” Oak 1 core
Conversely, well preserved areas are physically
sound, usually lighter in colour and macrostructu-
Thin decayed outer layer
res such as growth rings remain clearly visible. Oak “Half-Beam” Oak 2 and large sound core
The distribution of differently degraded regions Very thin decayed outer
was used by Christensen 17 to develop a classifica- layer and large sound
Oak “Half-beam” Oak 3 core
tion system, based on the oak timbers of a Viking
wreck from Roskilde Fjord, Denmark. De Jong 18 Sound pockets surround-
ed by highly degraded
later developed this scheme for European oak Poplar “Half-Beam” Poplar 1 area
(Figure 1), which has become widely adopted by Small sound pockets,
conservators. Valuable archaeological material large highly degraded
can rarely be destructively sampled to reveal Poplar “Half-Beam” Poplar 2 areas

internal degradation patterns. Instead, conserva- Thin decayed outer layer

surrounding large sound
tors typically assess timbers by extracting small Poplar “Half-Beam” Poplar 3 core
diameter core samples, probing with needles and
measuring the moisture content of fragments. Thin decayed outer layer
Pine “Half-Beam” Pine 1 and large sound core
In this study, several oak timbers from the Tudor
warship Mary Rose 19,20 are destructively sampled Thin decayed outer layer
Pine “Half-Beam” Pine 2 and a large sound core
to reveal their internal degradation patterns in
Very thin decayed outer
cross-section. These are described and visually
layer and large sound
compared to measured U max distributions throug- Pine “Half-Beam” Pine 3 core
hout each cut timber slice. Statistical distributions
Table 1: Timbers used as source material.

30 Moisture Content of Archaeological Waterlogged Wood, e-PS, 2008, 5, 29-35

 © by M O R A N A RTD d.o.o.

2 Materials and Methods using a band saw. Each slice was further cut to
produce approximate 1cm × 1cm × 1cm cubes
Source material for investigations comprised oak (Figure 2). For each of these, a note was made of
( Quercus robur L. or Quercus petraea (Matt.) its visible physical state (well preserved or degra-
Liebl.), poplar ( Populus tremula L. or Populus ded) and its maximum moisture content determi-
nigra L.) and Scots pine ( Pinus sylvestris L.) tim- ned.
bers from the Tudor warship Mary Rose . Species
identifications were confirmed microscopically Cubes were numbered and weighed in the water-
using the keys of Schweingruber 21 and Hather. 22 logged state using a digital balance accurate to
0.001 g, then placed in an oven at 103 ± 2 °C for
Initial tests compared actual moisture content 48 h and reweighed. Maximum moisture content
values to maximum (U max ) values. Six, 2cm × 2cm values were calculated using the following
× 2cm samples were cut from homogeneously equation: 23
degraded regions of each source timber (Table 1),
three of which were immersed in water and placed U max %=((W wet -W dry ) / W dry )) 100%
cyclically under partial vacuum to expel any trap-
ped air. The moisture content of all six samples where, W wet = weight of fully waterlogged sample
was then determined by the oven dry method as and W dry = weight of sample after complete desic-
described below. Negligible differences were cation.
found between actual and maximum moisture con-
Calculated U max values were plotted on a two-
tent values, indicating timbers were fully waterlog-
dimensional representation of each slice to illu-
ged. All subsequent measurements were therefore
strate their distribution. Collective data ranges for
made without first placing samples under vacuum
all three timber slices of each species were arbit-
and all figures given should be considered maxi-
rarily divided into four bands (Figure 3), which
mum values.
were used to colour-code the profiles.
To illustrate U max distributions throughout individu-
Average maximum moisture content values were
al timbers, profiles were produced from one centi-
also calculated for six 2cm × 2cm × 2cm replicates
meter thick “slices”, cut transversely from timbers
of modern samples of oak ( Quercus robur L.), pop-
lar ( Populus alba L.) and pine ( Pinus sylvestris L.)
heartwood, which were artificially waterlogged
under vacuum, to allow direct comparisons.

3 Results and Discussion

3.1 Oak

All oak timber sections showed degradation pat-

terns corresponding to class III of de Jong’s sche-
me (Figure 1), in general agreement with previous
studies of the hull’s timbers. 8,24,25 Values within
the inner cores were close to that obtained for
modern waterlogged oak (Table 2). High moisture
content values were concentrated in outer, visibly
degraded regions, as reflected in the moisture
content profiles of each slice (Table 3).

Application of de Jong’s classification using maxi-

Figure 2: Timber section after cutting to produce individual 1 cm mum moisture content values highlighted a poten-
cubes. tial problem. Using mean data for each slice
(Table 2), as de Jong would have intended, did
Minimum U max Standard Deviation Maximum U max
give classifications consistent with the visual
aspect of his scheme (Class III, <185% moisture
content). The success of this method however,
Mean U max
relies on the use of just one, average moisture
content value, for the whole timber slice or samp-
Band 1 Band 2 Band 3 Band 4
le. Contradictory results might be seen where only
Figure 3: Arbitrary method used to divide the total range of mois- small samples are available for destructive testing,
ture content values for each species into four “bands”.

Moisture Content of Archaeological Waterlogged Wood, e-PS, 2008, 5, 29-35 31

www.e-PRESERVATIONScience.org

as is often the case. This is evident in the ranges

of individual values for the oak slices reported
here (Table 2), which cover two of the classes
defined by de Jong’s scheme (Figure 1). An impro-
ved numerical scheme would ideally incorporate
two different moisture content ranges, represen-
ting well preserved and degraded areas of timbers
respectively. By studying the statistical distribution
of values within the slices studied here, it might be
possible to identify ranges of data that can be
used to form such a scheme.
Oak 1 Oak 2 Oak 3 Combined
U max U max U max data

Minimum 74 72 77 72 Figure 4: Histogram showing distribution of collective maximum

moisture content data for all oak timber slices studied.
Maximum 209 365 303 365

Mean 97 118 127 108 Collective moisture content data from all three oak
timber slices are presented as a histogram in
S.D. 26 49 39 38
Figure 4. Most values fall within a narrow range
Variance 658 2434 1513 1416 (70% to 150%), and were derived from the large
Moisture content profile “bands” calculated from combined data sound cores of the sections (Band A). A second,
and Fig. 3
less well defined population with higher range
Band 1 <96 (151% to 365%), is made up of data from degraded
Band 2 96 - 121 outer layers (Band B). A potential classification
scheme based on these data ranges is given in
Band 3 122 - 146
Table 4.
Band 4 >146

Table 2: Statistical data and moisture content profile bands for

oak slices. U max obtained for modern waterlogged European oak 3.2 Poplar
( Quercus robur L.) = 102%. S.D. = standard deviation.
Mary Rose poplar timber slices (Table 1) revealed
Timber section Moisture content profile considerable variation in their moisture content
values (Table 5) and visible degradation patterns
(Table 6). Timber “Poplar 3” was best preserved,
with a large sound core where U max values were
generally close to that obtained for modern water-
Oak 1
logged poplar (Table 5).Timbers “Poplar 1” and
“Poplar 2” both showed small irregular pockets of
well preserved material, surrounded by larger,
highly degraded areas. Again, visible degradation
patterns of all slices were reflected in their moistu-
re content profiles (Table 6).
Oak 2
Clearly a different classification scheme is requi-
red for these timbers, though de Jong’s system
provides a convenient basis for this. His system is
easily modified to compare relative proportions of
well preserved material versus highly degraded,
Oak 3
as given below :

Table 3: Oak timber slices and their corresponding moisture con-

Class A – Well preserved material dominates.
tent profiles. Class B – Roughly equal proportions of well pre-
served and highly degraded material.
Degree of Degradation U max Class C – Highly degraded material dominates.
Well preserved ≤150%
This would classify timber “Poplar 3” as class A
Degraded >150%
and timbers “Poplar 1” and “Poplar 2” as Class C
Table 4: Proposed classification scheme for oak. (Table 6). Squirrell and Clarke 8 present data for a
core sample taken from timber “Poplar 2” ( Mary

32 Moisture Content of Archaeological Waterlogged Wood, e-PS, 2008, 5, 29-35

 © by M O R A N A RTD d.o.o.

Rose timber number MR81T353) in general agree-

ment with this assessment.

Collective moisture content data are presented as

a histogram in Figure 5. Even in well preserved
areas, values are dramatically higher than seen for
denser oak timbers, and clearly a separate classi-
fication system is required based on these data.
Though less distinct than was seen for oak, two
broad data groups are visible that might be used to
form a basic classification scheme, as given in
Table 7.

Poplar 1 Poplar 2 Poplar 3 Combined

U max U max U max data
Figure 5: Histogram showing distribution of collective maximum
Minimum 378 170 100 100 moisture content data for all poplar timber slices studied.

Maximum 988 1162 817 1162 Band A (160% to 400% moisture content) is made
Mean 777 689 304 563 up of data from the sound “pockets” of timbers
“Poplar 1” and “Poplar 2” and the sound core of
S.D. 137 263 161 258
“Poplar 3”. Band B (401% to 1180% moisture con-
Variance 18799 69026 25853 66712
tent) was derived from highly degraded areas.
Moisture content profile “bands” calculated from combined data
and Fig. 3

Band 1 <477 3.3 Pine

Band 2 447 - 648
Of the three Mary Rose pine timbers, “Pine 3” was
Band 3 649 - 821 best preserved, with a very thin outer decayed
Band 4 >821 layer and large sound core. U max values were
generally close to that of modern waterlogged pine
Table 5: Statistical data and moisture content profile bands for (Table 8). Timbers “Pine 1” and “Pine 2” also
poplar slices. U max obtained for modern waterlogged poplar
( Populus alba L.) = 254%. S.D. = standard deviation. showed large well preserved cores, though had
thicker decayed outer layers (Table 9).
Timber section Moisture content profile
Use of the visual component of de Jong’s classifi-
cation might be attempted for these timber sec-
Poplar 1

tions, though outer decayed layers were notice-

ably less uniform than given in his scheme (Figure
1). Again, visible degradation patterns throughout
each slice were reflected in their moisture content
profiles (Table 9).
Poplar 2

Pine 1 Pine 2 Pine 3 Combined

U max U max U max data

Minimum 196 153 94 94

Maximum 670 485 221 670

Mean 322 262 156 244

Poplar 3

S.D. 135 105 23 120

Variance 18358 10951 546 14470

Moisture content profile “bands” calculated from combined data
and Fig. 3

Band 1 <204
Table 6: Poplar timber slices and their corresponding moisture
content profiles. Band 2 204 - 283

Degree of Degradation U max Band 3 284 - 364

Well preserved ≤400% Band 4 >364

Degraded >400% Table 8: Statistical data and moisture content profile bands for
pine slices. U max obtained for modern waterlogged pine ( Pinus
Table 7: Proposed classification scheme for poplar. silvestris L.) = 178% . S.D. = standard deviation.

Moisture Content of Archaeological Waterlogged Wood, e-PS, 2008, 5, 29-35 33

www.e-PRESERVATIONScience.org

Timber section Moisture content profile 4 Conclusions

Moisture content profiles from all timber slices

reflected visible degradation patterns, supporting
the use of U max data as an indicator of preserva-
Pine 1 tion state. Incorporating measured values in to a
classification system requires care however, parti-
cularly if only small samples are available for
destructive testing. The problem arises from the
fact that very different states of degradation are
routinely found within individual timbers. A small
Pine 2 sample could be taken that is not in fact represen-
tative of the bulk of the timber. If a numerical clas-
sification system is to be employed, it should utili-
se two well defined ranges of values for “well pre-
served” and “highly degraded” areas respectively.
Even then, this data is better treated separately
Pine 3 from, or complimentary to, more robust classifica-
tions based on visual appearance or physical exa-
Table 9: Pine timber slices and their corresponding moisture con-
mination. U max ranges will be unique to individual
tent profiles. species and a generalised scheme suitable for all
timber types is not possible. This is clearly illu-
Degree of Degradation U max
strated in the differences between the three sche-
Well preserved ≤250% mes proposed here for oak, poplar and pine. A
Degraded >250% common observation of potential interest however,
is that the values dividing the bands for the three
Table 10: Proposed scheme for Scots pine. schemes are close to the average U max values for
modern material plus 50%, as shown below. Some
caution is needed however, until this relationship
has been tested for other timber species.

U max modern +50% Scheme limits

Oak 102% 153% 150%

Poplar 254% 381% 400%

Pine 178% 267% 250%

Visible degradation patterns clearly varied

between the three species studied here, but this
need not exclude the use of a generic visual or
physical classification scheme from being used.
De Jong’s scheme for European oak 18 can easily
Figure 6: Histogram showing distribution of collective maximum
moisture content data for all pine timber slices studied.
be modified to compare relative proportions of well
preserved versus degraded material, making it
In Figure 6, showing collective data for the pine applicable to all timber species. A timber could
timber slices, the limits of populations represen- thus be described as belonging to class A, B or C,
ting well preserved and degraded regions respec- with approximate moisture content range A (well
tively are again not as well defined as seen for preserved core or pockets) to B (highly decayed
oak. Two broad populations are visible however, areas). A summarised classification scheme, com-
the first derived from the sound cores (Band A, bining visual and numerical aspects, is given
94% to 250% moisture content) and the second below.
from the degraded outer zones (Band B, 251% to
670% moisture content), as given in Table 10. Class A - Well preserved material dominates.
Class B - Roughly equal proportions of well pre-
served and degraded material.
Class C - Degraded material dominates.

Where “well preserved” and “degraded” can be

defined by the maximum moisture content values
as follows:

34 Moisture Content of Archaeological Waterlogged Wood, e-PS, 2008, 5, 29-35

 © by M O R A N A RTD d.o.o.

7. R.W. Clarke, J.P. Squirrel, The Pilodyn - an instrument for

Well Preserved Degraded assessing the condition of waterlogged wooden objects , Stud.
Conserv., 1985, 30, 177-183.
Oak ≤150% >150%
8. J.P. Squirrel, R.W. Clarke, An investigation into the condition
Poplar ≤400% >400%
and conservation of the hull of the Mary Rose. Part 1: Assessment
Scots pine ≤250% >250% of the hull timbers , Stud. Conserv., 1987, 32, 153-162.

9. P.A. Schniewind, Physical and Mechanical Properties of

It is not intended that the lengthy assessment pro- Archaeological Wood , in: R.M. Rowell, R.J. Barbour, Eds.,
cedures described here are duplicated for other Archaeological Wood. Properties, Chemistry and Preservation ,
American Chemical Society, Washington D.C, 1990, 87-110.
archaeological timbers. Indeed, the destructive
sampling required is rarely possible for valuable 10. Y. Kohdzuma, K. Minato, Y. Katayama, Relationships between
some properties of waterlogged woods , Mokuzai Gakkaishi, 1996,
archaeological material. Conservators will typical- 42, 681-687.
ly assess timbers by probing the surface with
11. R.A. Blanchette, Biodeterioration of Archaeological Wood ,
needles, extracting small diameter core samples Biodeterioration Abstracts, 1995, 9, 113-127.
and obtaining U max measurements from fragments.
12. R.A. Blanchette, T. Nilsson, G. Daniel, A. Abad, Biological
Instead, it is intended that the data and classifica- degradation of wood , in: R.M. Rowell, R.J. Barber, Eds.,
tion schemes given, are used as reference mate- Archaeological Wood. Properties, Chemistry and Preservation ,
rial that describe typical degradation patterns and American Chemical Society, Washington D.C, 1990, 141-174.

U max ranges for archaeological oak, poplar and 13. R.A. Blanchette, P. Hoffmann, Degradation process in water-
pine timbers. Data can then be used to comple- logged archaeological wood , in: P. Hoffman, T. Daley, T. Grant,
Eds., Proceedings of the 5th ICOM Group on Wet Organic
ment and aid in the interpretation of findings obtai- Archaeological Materials Conference , ICOM, 1994, 111-142.
ned from less destructive sampling techniques.
14. P.D. Evans, W.B. Banks, Degradation of wood surfaces by
water , Holz Roh-Werkst., 1990, 48, 159-163.
It should be noted however, that data presented
here are for a limited group of timbers, all of the 15. P. Hoffmann, A.M. Jones, Structure and Degradation process
for waterlogged archaeological wood . in: R.M. Rowell, R.J.
same age and from just one burial site. A broader Barbour, Eds., Archaeological Wood. Properties, Chemistry and
sampling strategy should ensure the full range of Preservation , American Chemical Society, Washington D.C., 1990,
degradation states has been explored for each 35-65.

species, and further studies are underway at the 16. Y.S. Kim, Chemical characteristics of waterlogged archaeolo-
Mary Rose Trust, assessing timbers from a wider gical wood , Holzforschung, 1990, 44, 169-172.

range of burial environments. Other species, inclu- 17. B. Christensen, The conservation of waterlogged wood in the
ding English Ash ( Fraxinus excelsior ) and Elm National Museum of Denmark , The National Museum of Denmark,
Copenhagen, 1970.
( Ulmus procera ) are also being examined.
18. J. de Jong, Conservation Techniques for old archaeological
wood from shipwrecks found in the Netherlands , in: A.H. Walters,
Ed., Biodeterioration Investigation Techniques , London, 1977,
5 References 295-338.

1. P. Hoffmann , Chemical wood analysis as a means of characte- 19. M.H. Rule, The Mary Rose : The excavation and raising of
rising archaeological wood , in: D.W. Grattan, J.C. McCawley, Henry VIII’s flagship , Windward, Leicester, 1983.
Eds., Proceedings of the ICOM Group on Wet Organic
Archaeological Materials Conference, Ottawa 1981 , ICOM, 1982, 20. P. Marsden, Sealed by Time. The loss and recovery of the
69-72. Mary Rose , The Mary Rose Trust, Portsmouth, 2003.

2. D.W. Grattan, C. Mathias, Analysis of Waterlogged Wood : the 21. F.H. Schweingruber, Microscopic Wood Anatomy , Swiss
value of chemical analysis and other simple methods in evaluating Federal Institute of Forestry, Switzerland, 1990.
condition , Somerset Levels Papers, 1986, 12, 6-12. 22. J.G. Hather, The Identification of the Northern European
3. E.A. Mikolajchuk, N.G. Gerassimova, A.A. Leonovich, A.V. Woods. A guide for archaeologists and conservators , Archetype
Obolenskaya, I.Yu. Levdik, S.Yu Kazanskaya, Examination of Publications Ltd, Great Britain, 2000.
waterlogged archaeological oak wood , in: I.D. Macleod, Ed., 23. C. Skaar, Wood - Water Relations , Springer-Verlag, Berlin,
Conservation of wet wood and metal. Proceedings of the ICOM 1988.
conservation working groups on wet organic archaeological mate-
rials and metals, Fremantle, 1987, Western Australian Museum, 24. A.M. Jones, M.H Rule, E.B.G. Jones, Conservation of the tim-
Perth, 1989, 95-107. bers of the Tudor ship Mary Rose , in : S. Barry, D.R. Houghton,
G.G. Hewellyn, C. O’rea, Eds ., Biodeterioration VI , CAB and
4. R.J. Barbour, The condition and dimensional stabilisation of Biodeterioration Society, London, 1986, 354-362.
highly deteriorated hardwoods , in: R. Ramiere, M. Colardelle,
Eds., Waterlogged Wood : Study and Conservation. Proceedings 25. A.M. Jones, M.H. Rule, Preserving the wreck of the Mary
of the 2nd ICOM Waterlogged Working Group Conference, Rose , in: P. Hoffmann , Ed., Proceedings of the 4th ICOM Group on
Grenoble 1984 , GETBGE-CENG ORIS, 1985, 23-36. Wet Organic Archaeological Materials Conference , Bremerhaven,
ICOM, 1991, 25-47.
5. Y.S. Kim, Microscopical and chemical characteristics of archae-
ological waterlogged wood , Abstracts of Am. Chem. Soc., 1988,
196, 28.

6. K. Borgin, G. Tsoumis, C. Passialis, Density and Shrinkage of

Old Wood , Wood Sci. Technol., 1979, 9, 49-57.

Moisture Content of Archaeological Waterlogged Wood, e-PS, 2008, 5, 29-35 35