David Gunton's Hardwood Floors, hardwood flooring, parquet, marquetry and boards, especially wideboards, in oak, ash, maple, beech, walnut, cherry, and many other woods.
David Gunton's Hardwood Floors.
Grange Lane, Winsford,
Cheshire, CW7 2PS
Tel: +44 (0)1606 861 442
Fax: +44 (0)1606 861 445

David Menu


Go back to home page.

Hardwood Flooring.

Thick or Thin? Wide or narrow?

By David Gunton.:

Awarded 1999 ; The Worshipful Company of Carpenters Special Award in Recognition of Outstanding Achievement in the Restoration, Re-Creation, New Design and Quality of Craftsmanship at Windsor Castle.

Just like the public, hardwood flooring specialists across Europe are subject to prejudices and misconceptions about their own trade and about what can and cannot be done with wood. These lead them into making false claims and assumptions about matters which are incontrovertible in science and logic. One of the most contentious areas of argument rages around thickness, width and cut of timber. Some say that hardwood flooring must be thick, rigid, solid, and narrow. Others insist upon thin, flexible and as wide as you like. Both sides are right - but each only in part.

Timber performance / behaviour.

Timber is hydroscopic. It gives off and takes up water by wetting with free water, and by absorbtion of water vapour from the atmosphere. Timber species vary very considerably in their hydroscopicity and their consequent propensity to associated expansion and contraction. Not only do species vary but the 'cut' off the log affects the degree of expansion and contraction.

Quarter sawn timber expands and contracts significantly less than tangentially sawn timber. A 1" thick tangentially sawn board (crown cut ) will have a different co-efficient of expansion on one face to the other. The density of the timber affects the speed with which it will dry. Some very dense tropical timbers can be difficult to dry even when machined very thin. Veneer will remain stable if securely bonded to a stable substrate and only deterioration of the adhesive will allow it to expand or contract. The Romans used dried oak wedges driven into holes, drilled by hand, then wetted, to split enormous slabs of marble from the quarry face. Warm humid weather causes more expansion in timber than cold rainy days. Warm dry desert atmospheres desiccates timber to the point of cellular collapse, but, equally, cold dry atmospheres can also create excessive dryness. Sailors of timber decked ships swab the decks when idle, not to provide work for idle hands, but to prevent the timber from shrinking and, later, letting in water when the weather cut up rough.

The point of the above list of apparently unrelated facts is to demonstrate that timber and moisture have a close relationship, that there can be slow, but powerful, even irresistible, reactions between them, that heat and cold have relevance, but less so than humidity.

i. Flying in the face of all logic, reason and evidence, the perception persists, that 'thicker is better' for flooring. This may hold some truth for barns, stables and industrial floors. For domestic, commercial or office use, which is to say, decorative hardwood floors, it is false. At Windsor Castle, the decorative marquetry floor in the Crimson Drawing Room, which the writer recreated after the fire, was only 2 to 3mm thick of veneer laid over a stable substrate. This veneer floor had lasted for some 140 years and had been in good condition prior to the fire. Some of the sections, cut from very unstable timbers such as sycamore, were 200mm wide. Even after the fire, some of those sections had retained their integrity and bond to the base despite being boiled continuously for 3 days! The marquetry borders to the floors in the State Rooms at Buckingham Palace are similarly constructed and have withstood considerable use and not a little abuse since the building was erected. At UMIST in Manchester the Victorian pitch pine block floors were laid at approximately 60mm thick. They distorted and wore badly and unevenly simply due to their excessive thickness.

ii. Take oak as an example. Throughout the ages it has been used in an enormous variety of sizes and qualities for just about everything in the built environment. It will continue to be used for the same purposes for the foreseeable future. Our understanding of its properties and behaviour are not new. The Romans understood it just as well as we do today, but perhaps not so precisely scientifically. In practice todays craftsmen and customers have lost some understanding of the timber because science and automation have separated us from the natural and we have begun to confuse the real with the hypereal. As an exaggerated example, take a 300 x 300 x 6000mm beam. No sawmiller will attempt to kiln dry this wood. Beams like this are sold 'green', meaning wet, and allowed to dry in service, or they are air dried for several years. They are not kilned. As the beam dries it splits. Even under the most carefully controlled conditions the beam splits. It splits because the outside dries faster than the inside. Even under very slow drying conditions the moisture from the inside core cannot migrate through osmosis to the outside as fast as it evaporates from the outside - unless the outside is kept wet all the time - in which case the beam never dries! (The Vasa at Stockholm and the Mary Rose are sprayed continuously with water to prevent them drying out and their timbers collapsing.) Even if the drying process was slowed to 100 years - the beam still splits because the coefficient of shrinkage / expansion is different for each layer of the annual rings. It is much greater towards the exterior of the log than at the interior. Therefore as the log dries a shrinking exterior is trying to fit around a nearly static interior. The only way to stop the exterior splitting is to drill a decent sized circular core (100mm diameter?) out of the center of the log.

iii. All this is true for oak boards. The thicker they are, the more difficult it is to drive the moisture from the core, and in doing so the more distortion is caused. Thus a board that is ventilated on one side but not on the other will dry - or be wetted - on one face but not on the other. Thus one face will shrink - or expand - whilst the other remains static. Unless that expansion is resisted somehow, the board will curve out towards the wetted side. Relatively minor changes in moisture ( 1 or 2 degrees change in the range of 8 to 12 % a.m.c) will produce an effect. Rapid changes cause greater curvature than slow changes. This is because there is a moisture gradient through the timber. It takes time for a 2 degree decrease (or increase) in moisture to equalize through the timber. Thus the thicker the timber, the longer it takes to equalize through from one face to another.

iv. As a general rule expansion or contraction in timber is irresistible. However, there are always exceptions to the rule. To draw an analogy; if you take an elastic band and wind it once around your finger with gentle tension, it does not hurt and the blood flow is not impeded. With the same gentle tension, wind it around 10 times. Now you can feel fairly strong pressure and the finger begins to go red. Wind another 10 times and the pain kicks in as the finger turns blue and swells. Leave it there for long enough and the end of you finger will die and drop off.

v. If you take a half millimeter thick leaf of oak veneer and bond it to a stable back ground with a rigid waterproof adhesive it will allow any amount of wetting but will be unable to expand because the cumulative strength of the few cells thickness is unable to break the bond with the adhesive. If you applied it wet, once the adhesive dried, no amount of drying will make it shrink off the background - short of burning it. In the attempts to expand or shrink, the cellular structure is either compressing or stretching.

vi. On the other hand, if you securely bond a piece of oak 300mm wide and 20mm thick to a stable background, then cause it to shrink or expand, the combined strength of the cells not actually in contact with the adhesive will cause the board to tear itself away from the adhesive, even tearing itself off the cells bonded to the adhesive.

vii. In addition to the complications of expansion and contraction and moisture gradients, there is the variations in coefficients of expansion between different cuts of the timber. The circular bands of the annual rings are the most commonly observed cellular structure of the wood. However, in addition to these there are the much less apparent medullary rays. These fine webs of cells radiate from the heart of the log to the exterior. Cut a board off the exterior of the log and you will have 'tangentially cut' or 'crown cut' board. This is the most unstable, the most prone to expansion and contraction of all the cuts of timber. It is the most prone to cupping and bowing as well because the difference between the coefficient of expansion of the outer face of the board and the inner face is significant. The appearance of the board is much loved by the American market and appears somewhat alien in European buildings.

viii. On the other hand, cut a board by slicing the log through its heart and then taking a board from that cut face and you have the most stable of all the cuts, a 'quarter sawn board'. In oak, this will show the medullary ray figuring. This is the cut most sought after through the ages by craftsmen making the finest furniture, paneling and flooring. It is not merely beautiful. Its stability is derived from the binding effect that the medullary rays have on the annual rings. The cells off the medullary act like staples binding the successive annual rings together. They will not allow the annual rings to expand much. The coefficient of expansion of a quarter sawn board is approximately that of a tangentially sawn board. However, quarter sawn timber is the most costly.

ix. Thus, it follows that, to produce a floor of wide boards which is very stable; it should be made of the thinnest timber possible commensurate with durability; the most stable cut of the timber commensurate with ones preference for appearance and the depth of ones pocket; housed in a controlled environment;

Use thin timber.

Veneer on furniture remains stable in widely variable conditions because it is bonded rigidly to the substrate. It is forced to compress rather than expand and to stretch rather than shrink. In hardwood flooring this property of flexibility is borrowed from the plywood bonded to screed construction and transferred by the rigid adhesive to the parquet fitted above. It has been found that 10mm thickness works adequately. This thickness satisfies the customer that he is getting value for money and a floor that will last. It is similar, if not a little thicker, than that thickness of wood above the tongue and groove of a 20mm thick board. That is the 'wear thickness'. Any floor which is worn or sanded down to the tongue and groove is effectively worn out. All the thickness below the tongue and groove is effectively wasted. It is there only to provide stiffness. Thinner than 10mm flooring would perform more reliably over under-floor heating under wider variations in atmospheric conditions. The wider the section, the thinner it should be to provide the greatest likelihood of dimensional stability. Note that thin veneers of several feet wide remain perfectly stable on furniture and plywood. Timber thicker than 10mm provides a rigidity and steeper moisture gradient which is counterproductive. The extra thickness is not required for wear - since, in practice, any floor which has lost even 3mm of its thickness is probably in a state of abuse and is worn out. The thickness required for rigidity can be provided by other stable material such as plywood. The Exceptions to the Rule. There are always exceptions to the rule. This company has successfully carried out very many exceptions to the rules including an installation of boards of Brobdignagian proportions, being up to 685mm wide (27 inches) by 7.3 metres (24 feet) long in 'country' grade, through and through sawn English oak at 32mm thickness over a piped water under-floor heating system. The installation is in a Grade II listed building set at the seashore. The floor is so low lying that if it were not for the bunds thrown up to keep out the sea, it would flood at spring tides. Yet, the boards have not changed dimension at all since installation. The stability has been achieved by ruthless adherence to all the principles set out in this essay - except those of choice of grade of timber, cut of timber, width and thickness. However, the client was warned that the boards will move if they do not maintain the evenness of the environmental conditions and that, in practice, they must expect some movement. In fact, in this case this movement in the boards is both desired and anticipated, since it will be the movement over many years which will improve the appearance of this floor which is constructed in an ancient tradition. Conclusion Hardwood flooring is susceptible to moisture, whether it be by loss of moisture, causing shrinkage, or by gain of moisture, causing expansion. This cannot be overemphasized. Significant variations of heat and humidity create strains in the timber, which it or its fixings cannot always withstand. Therefore these variations should be minimized. Correct installation procedures, with suitable adhesives and lacquers, all carried out in accordance with sensible precautions, and ensuring a consistency of temperature and humidity in service, will combine to provide satisfactory, lasting and beautiful flooring. There is a partnership in the maintenance of the stability of the hardwood flooring between supplier, flooring contractor, builder, heating engineer and, most importantly, the customer. Providing the flooring supplier and contractor have prepared, supplied and fitted the flooring with close attendance to the specification, subsequent movement in the timber must not be laid at their door. They are not magicians, capable of transmuting nature's mutable organic produce into the immutable inorganic. For further information, please contact us. David GuntonSaturday, 15 September, 2001


1. Oak and Iron.:

2. Underfloor Heating.:

3. Thickness of surface.: