What’s known about homegrown? (Hardwood edition)

Dan and I recently wrote a conference paper which Dan presented at the 8^th Hardwood Processing Conference in Coimbra: Hardwoods in the United Kingdom – Considerations When Looking to Future Planting and Future Value Chains. In the paper we explain which information on UK hardwoods is available, and where it comes from, and I just wanted to give a bit more detail on the origin of  mechanical property data of UK hardwood species.

We often say that not a lot of testing has been done on UK hardwoods, so we don’t know a lot about them. This is, however, not completely true and does not do the (sometimes massive) test programmes of the past ~100 years proper justice. We are also often pointing to “Lavers” (Gwendoline M. Lavers, The Strength Properties of Timber, first published 1967) as the only data that we do have, and we like to point out that these data are based on relatively little material. We also highlight that they are now quite old and tree properties might have changed since the testing was done (due to changes in forest management, climate etc). But where exactly do the Lavers data come from? Is there really no other data on home-grown hardwoods? And how can we compare historic results to the tests we are doing now?

Very early timber testing

The testing of timber mechanical properties has a long history, but only a few of the old documents are easily accessible today. These serve to show that there has been an interest in understanding timber mechanical properties for a long time, and to show the methods that were used in the early days of testing. Although no common test standard existed, three-point bending tests were commonly performed, and specimens often had a depth of 1 to 3 inches with widely-varying spans (from 4 to 72 inches in the research reviewed below). Compression tests parallel to grain were regarded as one of the most important tests, and often testing was done on specimens of 2×2 inch cross section, but different sections were also used and lengths varied (mostly between 6 and 8 inches). Some research included tensile [1] or shear tests [2], and others tested timbers for special uses, e.g. as wheel spokes [3].

Another factor influencing mechanical properties is the moisture content, and not all of the historic studies specify the moisture content at all, let alone measure it, although some differentiate between green and “seasoned” (and some studies give helpful hints like “cut from a piece of Oak timber that had been four years in store – it was not even then well seasoned” [4]).

A more detailed look

Already in 1670, Evelyn, in his famous publication Sylva, describes several experiments conducted between March 1663 and July 1664 to compare the strength of timbers, e.g.:

The test methods at the time were rather crude – they involved loading the timber with weight in the centre until it broke. The weight and dimensions of the timber as well as the breaking weight were recorded. The method is principally the same as the three-point bending test which is still being used (mostly for small-clear testing) today, only stiffness was not measured, spans were not reported, and sizes were variable. This was also a time before we had settled on proper standards for what an inch and a ounce actually is. It should take a few hundred years before this practice changed…

I did not dive too deeply into the very early history of timber testing, and so the next record I found comes more than 150 years later: Barlow of the Royal Military Academy contributed to our understanding of the theory behind material properties:  In his 1826 Essay on the Strength and Stress of Timber he performed bending tests with two fixed ends as well as only one fixed end, and developed equations to calculate strength and stiffness. He also gives results of his tests on home-grown oak, ash, beech and elm – Next to the bending tests (that use two different cross-sections and varying lengths), he performed tensile tests on turned specimens with a smallest cross section of ¼ inch.

From here on, MOE (Modulus of Elasticity) will be used in this blog post to refer to bending stiffness, and MOR (Modulus of Rupture) will refer to bending strength. Specific gravity refers to the density compared to that of water, but it is worth noting that this is, even today, often ambiguous. In the case of wood it can be used to mean density calculated from oven dry mass and green volume – which is easy to measure, but not a density that the wood actually ever is.

Laslett, Timber Inspector to the Admiralty, published some tests results on English oak in 1875. Although he uses quite different test methods than we are used to, his methods are well described and some results comparable to more modern results. However, these early studies do not really give reliable account of the moisture content, and even where Laslett states that “well-conditioned” timber was tested, high specific gravities and low MOEs in some of his results indicate this might not have been the case. Interestingly, Laslett performed experiments on the influence of test set-ups on the mechanical properties of oak, experimenting with the span and direction of loading in three-point bending, and the cross section and length of compression specimens. An undated report, probably published in the early 1900s [5], summarises testing that has been done between 1884 and 1909 at the R.C.D. (without explanation of this abbreviation). Again, the results are rather questionable in terms of moisture content, and for most experiments a density or specific gravity is not given either. Often only MOR and the size of specimens is documented. For some specimens of oak, ash, beech and English elm, bending and tensile properties as well as specific gravity are available. Impact strength was determined with variable weights, so that the results are hard to relate to more modern ones.

Some extremely well-documented testing has apparently been carried out at the City & Guilds (Engineering) College London in 1918-19 [6], but the records at the National Archives do not include test results (the last page of the report contains a hand-written note: “Diagrams in map cabinet” – should anyone know where this map cabinet is located, I would be intrigued to see these records!) Involved in this research were Prof. W. E. Dalby of the City and Guilds College, Mr. Robinson (presumably the Mr. Robinson involved in timber testing for this report and mentioned to be working at Manchester University and later at the Department of Aircraft Construction), Mr. Hugh Davies, and Prof. Groom. Professor Groom is a well-known name in forest and timber research at that time, and in a 1920 statement of the Forestry Commission, he is mentioned to have provided some data on home-grown timbers, which are summarised in this publication. It seems this summary reflects the data that have been collected according to the methods described in the 1918 and 1919 reports, as the timeframe, author, approximate specimen number, and most of the species (including obscure ones like “cluster pine”) match between both publications. We can assume, therefore, that the results relate to three-point-bending tests of specimens with 3×2 inch cross-section and 60-inch span, with a loading rate of 100 lb/min. The results were corrected to the target cross-section size and span, where the actual conditions deviated. During the bending test, the deflection was continuously recorded until the limit of proportionality, and the MOE was taken from the load-deflection diagram. Compression tests were done on 2x2x3 inch specimens and shear strength was tested on surfaces of 4 square inch.

The Technical Department of Aircraft Production published a report titled “Tests on Home-grown Timber” in 1918. The report contains some data on birch, wych elm, poplar, willow and sycamore tested in bending, tension and compression. The test methods are not included in the report, but it is mentioned that tests were carried out by Professor H. Beare and Sub-Lieut. Plugge. Professor Beare is likely Thomas Hudson Beare, who at the time was appointed as Regius Professor of Engineering at Edinburgh University. Not many works of his on the topic of timber testing can be found, but he published a paper in the Transactions of the Royal Scottish Arboricultural Society in 1906, in which he describes how timber testing was generally carried out at that time. For bending properties, he only mentions full-size tests, but the 1918 results are obviously small clear ones (as the MOR is high). Other reports published by Beare or the University of Edinburgh at that time could not be found, and so it is not clear which methods the professor preferred at time of writing the report. Since the testing was carried out for the department of munitions, it seems likely that the same methods have been used that were shortly afterwards described in a report by the Aeronautical Research Committee (1920) – compression tests on bobbin shaped specimens of 1 1/8 inch diameter and four-point bending tests on specimens 2x1x40 inch, as per Air Board Specification 2 V. 1 [7], but this remains speculation.

The 1920 report itself includes some results on timber testing, however. The data are based on test reports by Major A. Robertson of the Materials Section of the Royal Aircraft Establishment (RAE). Unfortunately, it is not exactly clear from this summary which timbers have actually been grown in the UK. A moisture content at time of test is also not reported. Only elastic properties are given for the bending tests.

The FPRL

Testing of home-grown timber for the systematic property categorisation really took up speed with the formation of the Forest Products Research Board, under the Department for Scientific and Industrial Research, in 1921. The board appointed five commissions on different lines of investigation that should establish a scientific understanding of wood and wood processing, and bring more home-grown and empire timbers into efficient uses. The initial committees were:

• Identification and structure
• Timber testing and the physical and mechanical properties of wood
• Seasoning and the general water relations of timber
• Preservatives of timber and the study of losses due to decay
• Extractives

Over the years the names and responsibilities of the different sections would change several times, but the principal lines of investigation remained the same for quite some time. In 1925, the Forest Products Research Laboratory (FPRL) was established in the facilities of the RAE in Farnborough, and only one year later the FPRL was moved to the new purposely-built facility at Princes Risborough nearby.

At this time, the Timber Mechanics section started “Project 1” on the physical and mechanical properties of different timbers. While the FPRL summary reports of the time recognise that testing of structural-sized members, next to small clear testing, is important for characterising design properties of timbers, Project 1 was focused on small-clear data to compare different species and so assess their potential for different uses. Some full-size testing has been carried out in the first half of the 20^th century, but this was mostly on imported softwoods, as it was assumed that the properties of home-grown timber could be modelled from clear wood properties and the effect of defects.

Project 1

On the bright side, we do get a lot of small clear data on home-grown and empire timbers from the early years of FPRL research (apparently more than 20,000 tests have been done in 1928 – see below). But “small clear” testing of 1928 is not necessarily the same “small clear” testing that is used today, and it is important to understand how test methods and standards evolved over time to be able to somewhat compare historic results to modern data.

Before 1929 no standard existed, so the tests described above, carried out before the Princes Risborough days, vary in specimen size, shapes and spans. At Princes Risborough, the test methods were formalised for the first time. Test methods  for specimens of 2-inch cross section, as well as sampling methods, are described in the first report published within Project 1: Forest Products Research, Project 1 Mechanical and Physical Properties of Timbers.

The test methods were the same as the ones that had been used at the US Forest Products Laboratory, Madison, since ca. 1902, and these were also commonly used in other countries, including Canada, India, New Zealand and some European countries. In Britain they were formalised as the basis for testing as British standard BS 373 in its first edition of 1929.

The 1929 version of BS 373 is almost identical to the abovementioned 1928 report. In terms of sampling, the standard refers to the same report, which specifies that at least five trees per species and locality should be sampled, and these should be “representative of the average dominant trees of a species in the locality”. Bending specimens are taken in pairs along the North-South and East-West axis of the stem, covering the whole range of radial positions. Specimens for testing at different moisture contents are also taken from different heights, depending on tree diameter. Specimens for most other tests are cut from the bending specimens after these have been tested. This seems like a very robust sampling strategy, and is actually quite close to what we are doing nowadays, when we want to screen properties of a species.

Importantly, the standard specifies that the moisture content at time of testing needs to be determined, and that the standard moisture content for seasoned timber is 12%.

The standard usually recommends specimen sizes with two-inch cross section. However, it also allows the use of smaller specimens for some tests, and deviations from the standard loading rates, which means that test results might still not be completely comparable. It is likely, however, that standard conditions have been adhered to at Princes Risborough – most reports just specify that “standard tests” were carried out.

Some of the test methods published in the standard had already been under critique by researchers of the Aeronautical Research Committee, and in the same year BS 373 was published for the first time a Committee was appointed to review the test methods. In 1934 the Committee on the Mechanical Testing of Timber proposed some revisions, that were reflected in the second version of BS 373 of 1938. The standard introduces the tension test parallel to grain and the compression perpendicular to grain test on two-inch cubes that is still common today. It also opts for turned specimens for tests of compression strength parallel to grain and tension strength perpendicular to grain. The 1934 report mentions that test results for compression strength parallel to grain in the turned, “bobbin-shaped”, specimens beech are “appreciably higher” (both in green and dry specimens) compared to the tests on prismatic specimens. The results in turned pine specimens tested in a “plunger apparatus” (see below) are 8% and 13% higher in green and dry specimens respectively. The variation of length between 4 and 8 inches in specimens with rectangular 2×2 inch cross section did not affect results.

With the 1957 version of BS 373, the “bobbin” method for compression strength parallel to grain disappears again, and nowadays only prismatic specimens are common. This means that results for compression strength collected between 1938 and 1957 might be some 10% higher than what we would expect, if the bobbin-shaped specimen was used for testing.

With the scarcity of timber during and after the second world war, using smaller test specimens for characterising species properties becomes desirable in Europe, and so the 2-centimetre standard for small clear testing is agreed as an alternative international standard in 1948. From 1949, it also becomes more common in Britain, and the 1957 revision of BS 373 recognises this change and adds methods for testing specimens of two-by-two-centimetre cross section. This version of the standard is still in use today, even though the now published version from the 1990s has some typos. Other national and international standards for small clear testing of timber exist in parallel.

The standard specimen sizes and loading rates of the mechanical tests in all three versions of BS 373 are compared below:

Which methods were used for the Lavers results?

The first edition of Lavers’ The Strength Properties of Timber was published in 1967, so well after the latest revision of BS 373. At first glance it is therefore not surprising that it contains data of small clear tests on 2-centimetre specimens. However, much of these data are actually older than BS 373:1957 and have originally been collected on 2-inch specimens. Some of the data in Lavers have been converted to the 2-centimetre standard, using conversion factors given in the appendix. Other differences in the test standards have also been somehow adjusted, such as the weight of the hammer in the impact bending test, but how these factors were established is less clear. Lavers also gives values for compression strength perpendicular to grain that are not directly measured but calculated from side hardness, giving an equation of unknown origin. These considerations give rise to several questions: When and how were the data on UK hardwoods originally collected? And how do the various conversions affect the results? Time for a bit more detective work.

The home-grown hardwoods we find in Lavers are:

Alder (from 8 trees)

Ash (from 67 trees)

Beech (from 36 trees)

Birch (from 8 trees)

Cherry (from 8 trees)

Dutch elm (from 18 trees), English elm (from 16 trees) and Wych elm (from 12 trees)

Hornbeam (from 12 trees)

Lime (from 2 trees)

Oak (from 21 trees), red oak (from 8 trees) and Turkey oak (from 6 trees)

Plane (from 1 tree)

Poplar (from 24 trees of two species)

Sweet chestnut (from 13 trees)

Sycamore (from 6 trees)

Willows (from 35 trees of three species)

Most homegrown hardwood data we find in Lavers have previously been published in a 1953 publication The strength properties of timber and its 1960 revision. The 1960 revision combines the 1953 report with a 1955 publication that establishes test methods and strength properties of some timbers to the 2-cm standard. This testing was done on 29 hard- and twelve softwood species, and the factors used in Lavers for converting the 2-inch results to the 2-cm standard were established from these tests. However, no home-grown hardwoods were included in the testing, and Italian beech served as the only temperate hardwood species to confirm the conversion factors. The correlations between the two sizes in individual species is also not shown in the report, but apparently in a report I have not been able to track down [11], so that it is not possible to see how much relationships vary between species. Within the Building from England’s Woodlands project (BfEW), we are again testing timber in these two different sizes, to understand better how the results of historic testing relate to new data. A conference paper on three-point bending tests has already been published, and it seems that some of the ratios used for conversion by Lavers are not necessarily holding up for our homegrown hardwoods. We continue to do more testing on various species, to confirm these findings.

But back to the origin of the UK hardwood data: For most species the Lavers data can be traced back even further. The 1939 Handbook of home-grown Timbers contains the same data on beech, cherry, hornbeam, plane, grey poplar, sweet chestnut, sycamore and the three willow species. The handbook, however, does not give us any more information on sampling. It claims to have been using the 1938 version of BS 373 as a basis for testing. But given that the standard had only been published one year before this second edition of the handbook, it seems more likely that this is an inaccuracy in the text, and most testing was actually done according to the earlier version of BS 373. (Notably, the handbook also gives a description of the working qualities and treatability of different home-grown timbers. These properties have been studied intensively at FPRL, and although we cannot give much thought to these in the BfEW project, we do think this might be another important quality criterion for certain sectors, so it’s great to see this information printed).

We can narrow down the time when testing was done for some of the species, looking at further publications from FPRL. For oak, we even find a report on the sampling and testing that also contains information about the wood structure, drying, protection etc. The mechanical properties published in the 1936 report The Properties of home-grown Oak are identical to the Lavers data [12]. It does not give us much information about sampling or testing, but this report tells us the trees came from Bedgebury, Kent; Oakenhill in the Forest of Dean; Brandon, Suffolk and Walcot Park, Shropshire. There are no large differences in wood from the different localities.

Some other data, while not identical to Lavers, seem likely to have been included in the various summaries published since 1939. The 1929 Report of the Forest Products Research board describes that a staggering 26,726 tests have been carried out on small clears in 1928, including tests on home-grown oak, beech, sweet chestnut, black polar and elm. Individual reports are not easy to find, but one 1929 report [13] alone contains 12,540 data points (so almost half of the data). It includes data on bending strength and stiffness, impact bending, compression strength parallel and perpendicular to grain, hardness, shear strength, cleavage, tensile strength and shrinkage of nine hard- and softwood species. So even though a lot of data were collected, for each species relatively few specimens were tested in each test, as can be seen in the graph below:

For example, an average of 77 specimens were tested in bending for each species. This includes tests in green and in dry condition. For dry English elm only four specimens were tested. The material sampled also came from only a handful of trees per species (five trees each of beech, oak and common elm, nine of ash and ten of Dutch elm).

FPRL often published some of their test data, only to later add to them and re-evaluate results, so it seems highly likely that this testing is included in the Lavers data. One of the four consignments of oak that is recorded in the 1936 report mentioned above also shows results that are suspiciously close to the one published in the 1929 report for some of the values, while others differ: The bending strength and work in bending as well as specific gravity of green specimens matches in both reports, while the dry data do not. The number of trees, growth rings per inch and summerwood percentage also match, so that it seems likely it is in fact the same sample, but some data might have been re-evaluated, adjusted or rounded differently… On the other hand, the impact strength, or the height from which a 50 lb hammer dropped broke the specimens, is different in both green and seasoned conditions between both reports, and this measurement is nearly impossible to re-interpret, so the two samples might be independent. In this case the 1929 data would be an addition to what is shown in Lavers.

Given the data that we have found and additional data the FPRL reports are alluding to, it seems likely that most of the testing we find in Lavers has been done in the very early years of FPRL research, before publication of the second version of BS 373. This means that compression strength parallel to grain is comparable to today’s standard, while compression strength perpendicular to grain results do not directly translate (anyway, no compression perpendicular to grain tests have surfaced in any additional publications so the values in Lavers remain a mystery. The good news is that the equation for calculating compression strength perpendicular to grain from side hardness actually seems to hold up really well!). Bending and impact tests were done on 2-inch specimens. Knowing this, the Lavers data can be compared with other historic data (and our new data from BfEW).

What other data are out there?

FPRL did not summarise absolutely all the data they had in their handbook of home-grown timbers or later publications. They investigated the kiln and seasoning properties of various species between ca. 1925-50 as part of “Project 5”, which involved mechanical testing of green, air-dry and kiln-dry timber. Two reports on Dutch elm [14] were published in 1931 and 1932. The test methods are very similar to the 1929 “Project 1” standard procedure, but strangely the results do not seem to be included in Lavers – the two reports comprise results from 28 trees, while Lavers only reports on 18. The data on Wych elm [15] from two 1932 reports also seem to be different from Lavers, even though 12 trees were sampled for the second report (as in Lavers). The same doubt remains for oak: A 1930 report [16] within project 5 includes testing on three different consignments of oak, three trees from each. The 1936 report contains trees from four consignments with five or six trees each, but two of the consignments come from the same places as two of the consignments in the 1930 report (Bedgebury, Kent and Walcot, Shropshire). On the other hand, if the data from the 1936 report on oak is actually the same data as in the 1929 report on oak (as argued above) then the Project 5 data have to be independent, as it is specified that tests were finished in 1930, after the first report had been published. Also the fact that 2×1.75 inch cross sections were tested in Project 5, instead of 2×2 inch cross sections as the standard for mechanical property characterisation at the time, suggests that the Project 5 data are an addition to Lavers. Beech was also tested within Project 5, and specimens from 26 trees are reported on in a 1928 report [17]. Two reports on Black Italian poplar from 1930 and 1931 include testing of 29 trees total [18].

Around the same time as project 5, tests were performed for a company to investigate the effect of vacuum kilning on ash. Only one tree was sampled and tests performed at two slightly different moisture contents, and results are published in the 1930 report Mechanical Properties of kiln-seasoned Ash [19].

Several reports published in the late 1930s within “Project 18” investigate the relationship between wood properties and wood structure (specifically the presence of tension wood). For these reports, many tests on compression strength parallel to grain and toughness were carried out, and these results are not included in Lavers. The compression strength for these tests was measured on “bobbin shaped” specimens with a 0.5-inch cross section and 3.5-inch length. Toughness was measured in a pendulum test on specimens with a cross section of 5/8 x 5/8 inch and 10 in length. Testing was usually done on matched specimens in green and air-dry condition, and the dry results adjusted to 12% mc. Although quite a bit of data was collected for this project, only few of the reports give summary statistics of the mechanical properties. Some results are available for the properties of normal and tension wood in beech [5] and ash [21].

Other factors affecting mechanical properties were also studied, such as the effect of moisture, temperature and structural anomalies. These test series contain data also on tests under standard conditions, that are additions to the Lavers data. Small glimpses of data are contained in the reports:

• Mechanical Strength Tests of Home-grown Ash (1933) [22]
• The Mechanical Strength Properties of ‘Brown’ Oak (1934)
• Relations between strength properties in static bending as determined from centre- and from four-point loading (1939) [23]

Since testing done after ca. 1960 is also not generally included in the temperate hardwood data in Lavers, additional data is found in the 1966 report “The quality of Ash from different parts of Britain” [24], which contains results of “standard tests” carried out on 67 trees from eight localities in England, Wales and Scotland – doubling the tree number from Lavers! It is, unfortunately, not specified if the 2-inch or 2-cm standard was used, but the latter one was likely more common at the time. The report also gives standard deviations to the means of the bending, impact bending, compression parallel to grain, hardness and shear tests.

Research after 1972

In 1972, the FPRL was merged with the Building Research Station and Fire Research Station to form the Building Research Establishment (BRE). Around the same time, a general revision of the British strength grading system started with the introduction of the new visual grading standard BS 4973 for softwoods in 1973 (the 2007 version of the standard with amendment 1 from 2017 is still in use today). BS 4978 introduced a new grade system, and this, as well as new practices of probabilistic limit state design should slowly replace the old CP 112 method, which used basic stresses and reduction factors, and so the testing of full-sized timbers gained more and more importance.

Already from 1957, researchers had been concerned with the formulation of grading rules for homegrown hardwoods, and visual grading rules had been formulated by the Hardwood Grading Committee, which was formed of FPRL researchers and representatives of the Home Timber Merchants Association of Scotland, The English Joinery Manufacturers’ Association, The Federated Home Timber Merchants, The British Furniture Manufacturers Association, the Royal Institute of British Architects, the Timber Development Association and the Forestry Commission. The grading rules were published as FPRL Leaflet No. 52 in 1958 and later formalised as BS 5657 in 1980. But these grading rules were not yet linked to design values for any homegrown hardwoods, because testing had not been done to establish mechanical properties according to the new international standards for structural design.

It was only in the 1990s that the BRE conducted testing on full-size timbers of two homegrown hardwood species – oak and sweet chestnut. Very briefly summarised results of these tests are published in BRE Digest 445 “Advances in timber grading”. So far we have not been able to find any more detail on this testing, which the only currently valid strength grading assignments for UK hardwoods are based on.

Some bits of homegrown hardwood data have been produced here and there in the past 20 years, but the volume of data collection from the early days of FPRL has not been repeated so far. However, the Building from England’s Woodlands and An easier Route to Strength Grading Hardwoods projects are making a good run for collecting the most UK hardwood data in the last 100 years – if we count to the same metric as the “12,000 datapoints” report above (counting every density measurement) we have so far collected more than 8,700 datapoints! Data collection in these projects will foreseeably continue until September 2025, so watch this space for a big info drop.

We are also in the process of visualising all the historic data that has been described here. You will need to wait until next year for this, too.

---

[1] 1875, Laslett, Timber and Timber Trees

[2] 1918, Dalby, Research on the Strength of British Timbers Research. Report No. 1 (in DSIR 36/1154)

[3] R.C.D Research Report No. 5, Tests of Timber (in DSIR 36/1154)

[4] 1875, Laslett, Timber and Timber Trees

[5] R.C.D Research Report No. 5, Tests of Timber (in DSIR 36/1154)

[6] Research on the Strength of British Timbers Research. Report No. 1. and No. 2 (in DSIR 36/1154)

[7] The method is partly re-printed in the Jenkins report and is likely the same as published in the revised standard BS 3V 4:1929 Specification for aircraft material. Ash. (earlier versions are not available)

[8] This method is still common in America, see ASTM, D 245

[9] Referring to the smallest section of the bone-shaped specimen

[10] Ibid.

[11] FOORD, A. J. Correlation of Strength Data under Schemes Based on 2-in. and 2-cm Cross-section Specimens. F.P.R.L. Report. 1955

[12] To see the results one has to visit the National archives, as they have not been captured in the scan, unfortunately. The report is found as record AY 18/858

[13] Forest Products Research: Tests of some home-grown timber in their green and seasoned condition in AY 18/1109

[14] AY 18/1183 and AY 18/1184

[15] AY 18/1158 and AY 18/1159

[16] AY 18/1147

[17] AY 18/1124

[18] AY 18/1151 and AY 18/1152

[19] AY 18/1045

[20] Progress report 5: Beech: Part 3, Part 4 and Part 5 “The distribution, structure and properties of tension wood in beech” (in DSIR 36/1154)

[21] Progress report 3: Ash: Part 3 and Part 4 “The Occurrence of Tension Wood in Ash (Fraxinus excelsior)” (in DSIR 36/1154)

[22] AY 18/1098

[23] AY 18/523

[24] In AY 18/901