Factors to Consider When Choosing a Stretcher or Strainer

Back to Paintings Chapter List

Return to Introduction.

The information presented on the Paintings Conservation Wiki is the opinion of the contributors and does not imply endorsement or approval, or recommendation of any treatments, methods, or techniques described.

Compiler: Barbara A. Buckley

Stretcher as Artifact

Aesthetic and Historical Considerations

The remarkable success of canvas painting supports from the 16th century onward likely owes much to the development of lightweight stretchers. Since their introduction, the design and construction of stretchers¹ have continued to evolve. Early stretcher designs had fixed joints, often a mortise and tenon or lapped construction secured with wooden pegs or nails. Stretchers with expandable joints appeared by the mid-18 th century with the introduction of the key. Some of the early keyed designs were problematic: single-key constructions produced uneven tensions, and keys were easily lost from open notches. This led to the development of the two-key design with grooves to better secure the keys. The 19th century ushered in the age of commercially produced stretchers. Many new stretcher designs were patented and are therefore datable. While there were significant improvements in stretcher design, such as the introduction of beveled edges, the overall level of individual craftsmanship deteriorated. The significance of these evolutions in stretcher design is that the stretcher itself is a valuable source of information in regard to the painting it supports.

As early as 1971, a philosophy that looked at the stretcher as a valuable source of art historical information was already underway. During a conference titled “American Painting to 1776: A Reappraisal,” Eleanor S. Quandt (1971) made a strong case for the stretcher as an important artifact. She describes two general categories of evidence drawn upon in the examination and documentation of an artifact: that which may be separated from the painting and therefore lost to the scholar, and that which is incorporated into the physical structure of the picture:

The first category comprises original painting supports—fabrics or wood panels—and their auxiliary parts: strainers and stretchers, nails and tacks. These ordinarily need no special equipment to be studied or sampled, and they tell far more about the historic character of a painting than one may realize. Most vulnerable in this category are original strainers. These exist in numbers which are diminishing rapidly, so that documentation of remaining examples is now a matter of considerable urgency. (Quandt 1971, 346)

Quandt continues this discussion by stating that the importance of stretchers and strainers has been neglected. They have seldom received careful examination, and they have even less often been preserved as a unique part of the painting. Much of this article is devoted to the types of information one may gather from the careful study of a stretcher. Quandt argues that even the best photographic and written documentary evidence may be lost and that the type of information considered important in one time period may change in the next.

Richard Buck in his landmark “Stretcher Design, A Brief Preliminary Survey” also makes a case for the stretcher as artifact, stating that “Examination of stretcher design is useful because it can reveal some evidence as to the origin, either place or time, of a painting—or at least of its restoration. In either case, it can offer clues about the history of the painting” (Buck 1992, 46).

The construction, design, joinery, and wood type of an original stretcher are all important elements of the painting as a whole. They can be used not only to date a painting, but also to place the painting within an artist's oeuvre, such as illustrated by H. von Sonnenburg (1979) in his studies of Rubens. In addition to the design and construction of the stretcher, auxiliary materials—whether intentional or not—may also provide a valuable source of information. The stretcher may contain various kinds of labels, such as auction, framers', exhibition, museum, and dealers' labels. Labels may also contain information related to the sitter or subject matter, provenance, or instructions for care of a painting. They may contain inscriptions by the artist, original or subsequent owners, a dealer, or framer that relate to the painting's provenance, subject matter, or attribution. In some cases, alternative light sources may be used to make inscriptions more legible. The stretcher may contain custom stamps, wax seals, or accession/inventory numbers. It is important to note that the original placement of labels, stamps, and inscriptions may be important information as well. The means of affixing the canvas to the stretcher (such as tacks, nails, or staples) and their placement may be significant or unique. Less commonly, an original stretcher may incorporate artists' materials, such as those found on rare, surviving original stretchers belonging to works by Albert Pinkham Ryder (Grimm 1998). Excess paint and/or ground on a stretcher may provide information related to the artist's technique or choice of materials. Even modifications to a stretcher may provide valuable information, i.e. relating to the reformatting of a painting.

Some questions to consider in terms of historical considerations are

• Is the stretcher original?

• Has a nonoriginal stretcher been documented, considering that it may provide information about the painting's conservation history?

• Has the original stretcher been fully documented and/or photographed, keeping in mind that the information may be lost and that the type of information considered important may change from what we consider important today?

• Does the design or construction of the stretcher provide information about the origins of the painting, such as period, date, region, school or, in unique cases, possibly the artist or the colorman's firm?

• Does the stretcher incorporate valuable information important for provenance or attribution, such as labels, inscriptions, stamps, or signatures?

• Does the stretcher represent a unique example of type or construction, or does it represent an anomaly or change within an artist's oeuvre?

• Does the stretcher incorporate the original artist's materials?

• Is the means used to attach the canvas to the stretcher—such as tacks, nails, or staples—and their placement unique or significant?

• Can the stretcher and/or means of attachment be reused?

• If original or unique, can the stretcher be structurally modified to properly support the painting?

Discussion of the aesthetic considerations of a stretcher is a bit more ambiguous. While stretchers have been overlooked as less important components of a painting, they do represent some level of choice (or lack thereof). Some artists specifically selected their stretchers; in some schools of thought, conservators will opt for the use of age- or artist-appropriate materials, such as the choice of specific woods and/or construction in the fabrication of a replacement stretcher. Most contemporary conservators would opt to retain significant or unique stretchers and/or tacks, nails, etc., when possible.

Charlotte Seifen Ameringer

Submitted January 2001

Why Use/Reuse a Stretcher/Strainer

The historical integrity of the painting is an important consideration in the decision to reuse a stretcher or strainer. The original size and shape of the picture is indisputably indicated by an unaltered original secondary support. Even if the original secondary support requires some alteration (for example, the addition of a bead to hold the fabric away from the inner edge or the modification of a strainer to allow for expansion), the artist's choice of secondary support, whether deliberate or unconscious, may convey other valuable information, such as the artist's financial state, the presence or absence of studio assistants, location at the time the painting was created, or conformity/departure from a pattern of working practices in the particular painting at hand. What appears to be a meaningless scribble or random mark on the secondary support may, in the context of the entire oeuvre or the provenance, prove to be an important identifier.

We cannot expect to be omniscient regarding the present and future discoveries related to the secondary support; it is therefore an invaluable resource. When the original stretcher/strainer is removed from the composite structure of the picture, although an attempt may be made to retain the essential information by photography, written record/description, or physical storage of all or a portion of the secondary support, it is much less certain that these documents will remain with the picture than that the actual support structure will.

If the painting is on its original stretcher/strainer, is it structurally sound? Can it be modified to make it sound? Has the original secondary support been the source of structural weakness/deformities in the primary support, and can it be modified to correct the problem? Are the original dimensions irregular, therefore increasing the difficulty of replacing the secondary support satisfactorily? If the painting will not be lined and the treatment planned will retain the original creases of the tacking margin, the picture will have the best “fit” on the original stretcher/strainer.

If the painting is unstretched, did the artist use a stretcher/strainer or a planar surface (e.g., a wall) while painting? What is the artist's preference or usual practice for exhibition purposes? If the picture is to be framed, will it be necessary to have a (replaceable) secondary support element (a stretcher/strainer), which can be secured in the rabbet for traditional framing techniques?

Catherine Metzger

Submitted November 2001

When to Replace a Stretcher/Strainer

Is the original stretcher/strainer missing? Is the physical integrity of the picture so compromised by the stretcher/strainer that no method of modification could adequately ensure its continued function as a tensioning device for the primary support? Some examples of loss of function are fractured wooden members, splintered joints, and severe warping. If the stretcher/strainer is not original, it may be replaced simply to supply a stretcher/strainer with improved mechanical properties or of more desirable dimensions. If the picture was originally on a strainer, has a history of requiring adjustment to maintain a planar surface, and the strainer cannot be satisfactorily converted into a stretcher, a new secondary support may be chosen. The selection of the support should take into consideration the future travel and display criteria, as well as the probability that a professional will adjust the tension when necessary.

If the painting has been lined with heavy or stiff adhesives, had interlayers added between the original and lining fabrics, or has been adhered to a lining fabric that is much stiffer than the original fabric, the original stretcher/strainer may not be strong enough to keep the composite support in plane. A new stretcher/strainer is often needed following lining treatments that significantly alter the weight and/or mechanical properties of the support.

Catherine Metzger

Submitted November 2001

Stretcher Design

Profile of a Stretcher

A beveled stretcher profile that slopes continuously from the outer to the inner edge of the stock or a raised bead on the outer edge of rectangular stock raises the reverse of the fabric support away from the stretcher. This helps to reduce the formation of a microclimate between the stretcher and fabric that is different from the climate behind the rest of the support. This discrepancy causes the different regions of the support to expand or contract at different rates. A bead or bevel can also prevent the support from touching the inner edge of the stretcher due to vibration of the support during transit or contact with the stretcher if the support is slack and concave on the stretcher. Impact of the support against the stretcher and repeated formation and cycling of microclimates behind the stretcher result in breakage of the ground and paint films above the inner edges of the stretcher. Eventually this causes the support to form raised creases along the cracks, a phenomenon known to conservators as “stretcher creases.” Separating the support from the stretcher also prevents adhesion of the support to the stretcher when glue size or acrylic priming is applied and squeezes through the fabric's reverse. Intermittent areas where the support is adhered to the stretcher in this manner may result in a dimpled, puckered deformation of the support over the stretcher members.

A stretcher profile that does not adequately separate the support from the stretcher can be retrofitted with strips of beading attached to the outer edges of the stretcher's inner face with glue and brads (fig. 1). An historic stretcher can thus be reused and improved in its function without removing any wood (see section IV.A.1).

Early European and American strainers and stretchers tended to be rectangular in cross section, with some lightweight strainers on small paintings almost square in section.¹ By the early 19th century, the practice of chamfering or beveling the inside face of the strainer or stretcher arose.

FIGURE 1 Stretcher with applied bead

Courtesy of Mark Bockrath

The earliest chamfering treatments were done by hand after the stretcher was assembled by scraping away at the inner edge of the member with a spokeshave or file. These are easily recognized by the rough contours of the worked edges and by the fact that the chamfering stops short of the corners of the stretcher, resulting in an oval-shaped outline to the chamfering. Therefore, the deepest chamfering is generally at the middle of each stretcher member. By the second quarter of the 19th century, stock was milled to create a continuously beveled profile before the corner joints were cut and the stretcher was assembled. After the mid-19th century, stock was sometimes milled to create a raised bead on its outer edges. By 1882, the Pfleger Company in the United States patented the familiar stretcher profile with mitred mortise and tenon corner joints and a raised bead on both the front and reverse outer edges of the stretcher, thus making a symmetrical and reversible profile. This design was later produced by the Anco Company, Bay State Stretchers, and Fredrix, among other American manufacturers. It remained in several variations as one of the most popular stretcher profiles in 20th century America for small to medium-sized paintings and is still in wide use today.

Despite these developments in beveled profiles in 19th century stretcher manufacture, it is common to see many rectangular profiles with no raised beads or bevels on strainers and stretchers from the 19th and 20th centuries. Homemade stretchers and strainers for large contemporary paintings and some auxiliary supports for folk paintings of both the 19th and 20th centuries may exhibit such profiles. Lightweight rectangular strainers with pinned corners and a rectangular profile are frequently seen on small French paintings of the late 19th and early 20th centuries.²

Mark Bockrath

Submitted October 2006

When to Use a Cross-Member

When a painting is larger than 30 to 36 inches in its longest direction, it may become necessary to use a cross-member in the construction of its stretcher, regardless of the stretcher design and member profile, unless the members themselves are very heavy. This is necessary due to the possibility of the stretcher twisting out of plane from the stress of the stretched fabric on its corner joints. The stretcher members may also warp due to the tension of the fabric pulling inward at the centers of the members if no cross-bracing is utilized. The problem of stretcher twisting is, of course, greatly exacerbated by a flimsy, lightweight stretcher profile, weak corner joints, or a very strongly tensioned, densely woven support fabric. Increasing the tension on the fabric by keying out the corners can also cause the stretcher to torque as its corner joints weaken when they are opened up by keying.

The cross-member should be placed perpendicular to the long member of the stretcher, halfway across the member so as to bisect it. Placing the cross-member in this position greatly strengthens the stretcher, and on larger stretchers, multiple cross-members provide greater strength. If a single cross-member does not adequately strengthen the stretcher, a great increase in stability can be achieved by adding another cross-member that is perpendicular to and bisects the first cross-member. The cross-members are half-lapped together at their juncture; the joint here becomes extremely strong when it is glued and screwed together. On very large stretchers, it may be necessary to add several more cross-members in each direction, with more cross-members running parallel to the shorter side of the stretcher if it is considerably longer in one direction than the other.

In addition to providing resistance to twisting of the stretcher, cross-members provide greater ease in evenly tensioning the support when they are themselves keyable. A common way to provide for keyability of the cross-members is to make each of the joints of cross-members keyable to outer members, with mortise and tenon joints, for example. This allows for very precise keying of several parts of the painting. Keying out a cross-member can more efficiently tension the center of the painting, thereby reducing slackness and bulging of the support here than if corner joints only were expanded. Thus, only a slight opening of the cross-member joint can add more tension to the center of the support than a more exaggerated opening of the corner joints. This is because the center of the painting is farther away from the corners than the more centrally located cross-members. Indeed, in many cases, the corner joints would need to be opened so much that they would weaken or break before adequate tension could be provided for the center of the painting if no cross-members are present. Multiple cross-members can allow such precise keying that localized slackness or bulging in the support can often be eliminated with a great degree of specificity and control, with the ultimate goal of having all regions of the support in approximately even tension.

If the cross-member is not joined to the outer member with a mortise and tenon or other joint, a butt-joined cross-member can provide more strength to the stretcher if it incorporates dowels or screws through the stretcher and into the cross-member or is secured over the butt-joint on the stretcher reverse with mending plates and screws. Strength also increases as the thickness of the cross-members approaches that of the outer members. These methods of adding a butt-joined cross-member do not result in a keyable mechanism and therefore provide strength rather than the ability to adjust fabric tension. They can frequently be added to a stretcher with very little work and may flatten a twisted stretcher or strengthen a flimsy one.

The stock used for a cross-member should be shallower in depth than the stock used for the heavier outer members; when the reverse of the cross-member is flush with the reverse of the outer stretcher members, a gap between the cross-member and the reverse of the support fabric is thereby provided. This gap helps prevent the cross-member from pressing against the reverse of the support, with subsequent formation of stretcher creases and lines of cracked paint over the edges of the cross-member.

More substantial stretchers that are commonly used as replacement or new stretchers by conservators vary somewhat in the design of their cross-members, owing to the requirements of their different stock profiles, corner joinery, and keying mechanisms (personal communications, 2000). Firms that manufacture these stretchers vary in their estimate of how long a stretcher member can get before it requires a cross-member. Depending on the stretcher design and the weight of the stock used, this measurement may vary between 24 and 40 inches, with 30 to 36 inches being typical. Expansion-bolt stretchers employ cross-members that are attached to the outer stretcher members with stainless steel dowels. The same expansion-bolt mechanism used in the corners of the stretcher can be used for the cross-member joints. If multiple cross-members are perpendicular to each other, they are joined with half-lap joints. ICA spring stretcher designs employ the use of a cross-member when the stretcher exceeds 35 inches in length. The cross-members include aluminum tubing and redwood stock in combination, with the tubing running through the redwood stock at a right angle. The redwood stock is used to make the longer cross-member. Both cross-members are attached to the outer members with threaded bolts; the joints are expanded by means of springs that are tensioned by tightening nuts along the threaded bolt, repeating the construction of the corner joint mechanisms. More traditional stretcher designs with simple or mitered mortise and tenon corner joints utilize simple mortise and tenon joints to join the cross-members to the outer members. The cross-members are half-lapped together as necessary, and all joints are expandable with wooden keys.

Mark Bockrath

Submitted October 2006

Choice of Materials Used for Stretchers/Strainers

There are many factors to consider when choosing the material for a stretcher or strainer. The material should be lightweight, durable, and dimensionally stable relative to the painting, as well as affordable, readily available, and easy to work with. These qualities can be found in the many types of materials from which stretchers have been made, such as wood, metal, and plastic.

Materials

Wood

It is not surprising that wood has traditionally been used for stretchers and strainers, since it can be lightweight, durable, relatively stable (depending on wood species and cut), inexpensive, and easy to obtain. Although the history of stretchers and strainers has not been well documented, it is possible that the cheapest wood available or scrap pieces were used to construct the earliest auxiliary supports (ICOM 1960, 150). During the 19th century, it is documented that the cost of making stretchers was important to their manufacture. Patents from the 19th century demonstrate this concern: E. H. Collins's patent states his goal was “to furnish a device for stretching prepared canvas for artists' use, in the most simple and economic manner,” and A. Stempel's patent states, “the cost of manufacture is also lessened” (Katlan 1992, 33). The most common type of wood used for stretchers is from the conifer family. Some of the earliest examples of strainers that have survived are made from simple, unfinished, pinewood frames. Mantegna's Presentation in the Temple (c. 1460) has a simple pinewood frame with a panel inserted at the back. The canvas is nailed to the front (ICOM 1960, 151). In America during the 19th century, many artist suppliers, such as A. H. Abbott and Company, advertised “Pine Strips for Stretchers” in 1900 (Katlan 1992, 311) or stretchers “made in our Factory from well seasoned clear pine,” as proclaimed in an 1891 trade catalog from Consolidated Copying Company, Inc., Chicago, Illinois (Katlan 1992, 317).

When conservators search for a stretcher, wood is the most likely choice of material for reasons beyond the financial. A stretcher made of wood that is lightweight, strong, and does not warp is desired to provide the best support. A survey of various contemporary stretcher manufacturers yielded results about the types of wood used to make replacement stretchers. In general, a fine, even-grained, defect-free wood that was kiln-dried is the standard. The kiln drying allows the atmosphere to be controlled so the drying of the green wood is slow and even, minimizing the possibilities of warping and checking (Hoadley 1980, 72). The specific type of cut used to harvest the timber from the rough tree is not described in the literature. Radially cut wood is ideal since its dimensional change is less than that of tangentially cut wood (Hoadley 1980). Both softwoods and hardwoods are used to make the stretcher members, including pine, cedar, redwood, poplar, and basswood. These types of wood are chosen for their low tendency to warp. Sugar pine, white pine, cedars, and poplar all have a low tendency to warp during seasoning, and Douglas fir, other pines, and basswood have an intermediate tendency to warp during seasoning (Hoadley 1980, 80). Since many of these woods have very similar characteristics and some are easier to obtain due to their natural habitat, there does not appear to be one preferred species of wood for making stretchers. As an example, the Tate Gallery uses stretchers made from a softwood, Douglas fir, but they also recommend sugar pine from North America, and the keys are made from beech, a strong, dense hardwood (Booth 1989, 36). Keys can be made from any dense wood, including cherry and maple.

See the appendix that follows for further information on specific softwoods and hardwoods.

Metal

There are several important factors to consider when choosing whether or not to use a metal stretcher: the dimensional stability of the stretcher, its interaction with the environment, and the method of attachment of canvas to metal. The metal chosen to make the stretcher should be strong enough to support the weight of a stretched canvas of various dimensions so that the metal stretcher does not bend or sag under the weight of a large canvas. Fortunately, metal can be manufactured in sections that are perfectly straight and strong and can be fabricated in any desired length and width. Unlike wood, metal will not warp with changes in relative humidity (RH). What may be a concern with raised levels of RH is the possibility of a metal stretcher rusting; therefore, it is essential that the correct type of metal be chosen or that the metal be coated. How the canvas is attached to the stretcher may also affect how a metal stretcher can be used. Either the stretcher design must incorporate wood at the edges so the canvas can be tacked or stapled, or the stretcher needs a specially designed attachment, such as a clip. Another factor to consider is the aesthetics of a painting stretched on a metal stretcher. Since metal is not a traditional stretcher material, a 17th-century painting on a metal stretcher may not be aesthetically pleasing to conservators, museums, or clients.

Research has brought to light three types of metal stretchers. The first is a historical design from 1941 made of Monel metal or steel; the second and third are both currently available stretchers made of aluminum: a stretcher designed and produced by Franco Rigamonti and the Starofix Stretcher.

Monal metal/steel

F. C. Osborn's design (patent #2,244,473, June 3, 1941) was called a stretching frame. The structure consisted of two stretchers/frames that allowed the canvas to be stretched without permanently fastening it to the stretcher/frame. The outer frame would preferably be made of metal, such as Monel metal (an acid-resisting alloy of nickel, copper, iron, manganese, carbon, and silicon) or steel. The inner frame could be metal or wood. The structure would be light and strong, the same qualities desired in wood stretchers (Katlan 1992, 200–202).

Aluminum

The first aluminum stretcher with automatic tension was designed and produced by Franco Rigamonti in 1966. The stretcher is made of extruded aluminum channels that are mitered at the corners and held together by aluminum joints. The joints have screws and helical springs mounted on each end that can be tightened through a slot on the back of the stretcher. The canvas is attached with specially fabricated metallic canvas holders that fit into a channel on the reverse of the stretcher. The specific type of aluminum was not found in the literature, but it is likely an anodized aluminum to prevent corrosion.

The Starofix Stretcher, produced by Starofix North America and developed in 1984, is a continuous tension stretcher and is made of aluminum (specifically an alloy temper 6063-T5 anodic coated 20 mym). This anodized aluminum is corrosion resistant. To solve the problem of attaching the canvas, strips of wood have been fastened to the outside edge allowing the canvas to be tacked or stapled in place. This aluminum stretcher is lightweight and strong, not needing a crossbar at dimensions up to 72" x 72". This makes the stretcher much lighter than wood by comparison, and it does not distort, bend, or sag (Starofix North America brochure).

See section III.B.4.a for further information on these aluminum stretchers.

Plastic

Research to date has uncovered only one source for a stretcher made of plastic. Craft Cut Products of Santa Fe, New Mexico, introduced the WONDERBAR™ in the late 1990s, although the company has since discontinued production. The stretcher was made from a recycled, extruded, composite material that resembles sawdust or wood pulp in a hardened plastic matrix. This recycled wood material was advertised as having no moisture absorption and no toxic emissions and that the stretcher members would not bend, twist, or warp. The stretcher members were cut at the desired angle and assembled with plastic corner inserts that fit into a cavity (molded into the stretcher member) and that hold the stretcher members square. A second plastic mechanism can be added at the corners to allow for keying out of the canvas, one plane at a time. The WONDERBAR™ was a heavy-duty stretcher bar and was quite heavy in weight, especially if assembled into large stretchers. The manufacturers cited weight as one of the reasons for discontinuing it. Aesthetically, this stretcher was the color of wood, being made from wood pulp, but it did have a plastic appearance. As stated earlier, a modern-looking stretcher may not be visually acceptable on some types of paintings (or to some people).

Weight Considerations

The dimensions of a painting determine the ultimate size of the stretcher, so the larger the stretcher, the more weight is added to the stretched painting. A material that is lightweight is desirable when stretching large canvases so that they can be safely handled and mounted on the wall. Wood has proven to be a versatile stretcher material. Some woods, such as those listed below, are light enough that large stretchers can be made and handled. The introduction of metal stretchers has made larger stretchers even more lightweight and easier to handle and mount. The plastic WONDERBAR™ proved to be quite heavy and was only suitable for smaller dimensions.

Alexis Miller

Submitted November 2006

Choice of Materials Used for Keys and Other Joint-Adjusting Mechanisms

Introduction

The purpose of the stretcher key is to control the expansion of the corner joints of a stretcher (Chase and Hutt 1972, 13) in order to adjust the tension of the fabric mounted on the stretcher. Not surprisingly, information on stretcher keys is scarce: little attention has been paid to keys historically, and research on them has always been hampered by the difficulty of determining whether they are original to a painting and/or to its stretcher (see section III.A.2).

Background Information

Stretcher keys made their appearance with the advent of keyable stretchers, first mentioned in France in the mid-18th century (Quandt 1971, 349-350). (There is some debate as to whether expandable stretchers may have appeared as early as the late 17th century Katlan 1999–2000). These stretchers probably developed out of the 17th-century tradition of lacing canvases for paintings to strainers to apply tension to them by tightening the laces (Katlan 1992, 20). Leslie Carlyle mentions prepared canvas available on “plain or wedged frames” in the c. 1842 Winsor & Newton catalog (Carlyle 2001, 186), which implies expandable “frames” were well established by then. Katlan confirms that the term “stretcher” was probably an Americanization of the British “stretcher frame” or “canvas frame” (Katlan 1992, 24). In America, keys first appear in the late 18th century when keyable wooden stretchers of several kinds were being used concurrently with strainers. Single-keyed, wooden stretchers seem to be most prevalent on “naïve” paintings of the early 19th century, and they disappear by the 1850s in America (Katlan 1999-2000). Whether the single-keyed stretcher is a primitive form of the double-keyed stretcher and concurrent with it or is simply the earliest form of the expandable stretcher needs more research. In the 19th century, American inventiveness produced a plethora of designs to improve the traditional single- and double-keyed wooden stretcher, as a review of the patents dating 1849 through 1947 in Alexander Katlan's 1992 book reveals. These patents include many variations on stretching systems and many expandable corners with or without metal additions and with or without single or double keys as an adjunct to the systems.

A discussion of the materials used for keys is rarely part of a discussion of how wooden stretchers are made. A review of the stretcher-keying patents in Katlan's book reveals that none of the patents mention what woods had been or should be used for wooden stretcher keys (Katlan 1992, 16–260). Katlan suggests this was most probably intentionally left out to give the patent its broadest application (Katlan 1999–2000). A brief review of historical artists' manuals also finds no mention of the type of wood from which keys should be made.

Of interest is that the term “wedge” has been used interchangeably with “key” from early times in America and possibly also in Britain, although “key” is the more common term now in both Britain and America. The earliest American patents listed by Katlan use both terms, for example: “and the wedges, or keys, by which the canvas is to be stretched” from an 1849 patent (Katlan 1992, 68) and “avoiding the use of keys or wedges” from an 1866 patent (Katlan 1992, 72). Later in the 19th century, “wedge” predominates, although in an 1893 patent for an improved stretcher “key” appears again interchangeably with “wedge”: “Figure 1 is a perspective of my improved key…; Fig. 2 a plan view of the same showing the wedge B in section” (Katlan 1992, 152). Even today, both terms can appear: in a website ad for Upper Canada Stretchers® appears “keys, wedges are used…to expand the frame in very small increments…”.

A review of the historic American patents reveals the same complaints about wooden keys at that time as exist for conservators now. For example, the Shattuck patent of 1885 states, “These wedges…were very inconvenient in use, as they were quite likely to split in driving, and also to shrink and fall out in a short time” (Katlan 1992, 110).

Materials

Generally all recent sources agree that wooden keys should be made of some sort of hardwood. This is for the ease of cutting hardwood and for its ability to take hammer taps without splitting. Beechwood is recommended by Booth in his article on stretcher construction at the Tate Gallery (Booth 1989, 36); Wingrill, stretcher makers of Ontario, Canada, makes its Conservator's Stretcher's keys of maple (Wingrill 1999). Edgar Kuschan, a stretcher maker in the Washington D.C. area, likes hardwoods such as cherry and walnut because of their attractive appearance and often uses scraps from around his shop (Kuschan 1999). Simon Liu, maker of painting supports in New York City, generally uses good quality poplar, kiln dried, because the wood is of medium hardness. If the wood is too hard, it can be difficult to cut, and possibly the key would be difficult to insert into the stretcher. If the key needs to be very thin, Liu may choose a harder wood such as maple or cherry for strength (Liu 1999).

Alternatives to wooden keys for traditional stretchers have been available for some years, most noticeable recently in the form of plastic keys. These are usually disapproved of by conservators for their lack of aesthetic appeal, lack of similarity to the original, and tendency to slide out of their grooves. Information from one American manufacturer of artists' materials, Tara-Fredrix, formerly the E. H. Friedrich Co., indicated that they started producing plastic stretcher keys (which they call “pegs”) in the early 1970s when first manufacturing plastic stretcher members. The plastic keys could be provided in the form of a T-F (Tara-Fredrix brand) “peg tree” in which eight keys are attached by a thin peg to each side of a thin post and can be broken off for use. This avoided the use of the small polybags containing eight wooden keys that would be included in the back of a shrink-wrapped, pre-stretched canvas on a wooden stretcher (Tara-Fredrix 1999).

However, plexiglass keys (see section IV.L) have been proposed by a conservator to solve the problem of replacing very thin wooden keys (less than 3/32" thick). 1/16" thick plexiglass was used to create the keys and the edges were roughened with coarse sandpaper for tooth, although the recommendations were also made for dipping the keys in Lascaux™ Acrylic Adhesive 360HV before inserting them or tying them in place with nylon monofilament cord anchored to the corner of the stretcher to hold them in place (Cockerline 1991).

I could find no evidence of the use of metal to fabricate the traditional triangular keys. Most of the various metal joint-adjusting mechanisms that have been proposed over the last two centuries to deal with the shortcomings of stretchers adjustable only with wooden keys were integral parts of new stretcher designs, and most of them are no longer in use (see the American ones in the patents listed in Katlan's book). Information about those metal joint-adjusting mechanisms still in use can be found in other entries describing stretcher designs. However, some of the historic metal pieces are described as keys by their inventors or in the trade catalogs advertising them (“My invention consists of a metallic plate and wedge combined, forming a stretcher key to be applied to the miter-joints of frames used by artists” as stated by Aaron D. Shattuck in his February 13, 1883 patent). Refer to the patent designs in Katlan's book for further examples of metal corner key replacement systems (Katlan 1992).

One modern metal corner-adjusting mechanism that is presently available as a separate item attachable to any inner stretcher corner to expand it is a “tensor-bolt,” brought to my attention by Jay Krueger, Senior Conservator of Modern Paintings at the National Gallery in Washington, D.C. His example, available from a German firm, is a small, brass 4-piece item with an adjusting bolt. It can be screwed to the inner edges of a corner joint of a stretcher, then expanded by turning the bolt. A discussion on independent stretcher expansion bolts took place on the conservation DistList in October-November 1999 during an exchange on the availability of spring stretchers, and it was clear that such items could be, and probably have been, manufactured by independent machine shop workers for artists and conservators.

Form of the Key

The most thorough recent discussion on making wooden keys is in the 1989 article “Stretcher

Design: Problems and Solutions” by Peter Booth of the Tate Gallery:

The chosen angle for the keys will vary between about 14° for heavier stretchers and about 19° for smaller stretchers. The former angle will result from a pair of keys being cut from a slat of Beech with a length to width ratio of 4:1, and the 19° angled key from a slat of 3:1 ratio. Most Tate stretchers bear keys of about 16° cut from slats with a 3 1/2 :1 proportion. If the key angle is made significantly greater than 20° the act of tapping in a key will become less selective with an increased risk of both the joint members being forced away from the canvas centre. Care is taken to ensure that the key is of the same gauge as the key slot to minimize the risks of stretcher splitting. To reduce damage to the key, the key is cut with the key grain running parallel with the receiving angle of the key slot, which in turn will have been cut to the same angle as the key. (Booth 1989, 36)

Most agree that the average key angle should be around 16°. Edgar Kuschan also mentions the importance of the key's fit in its slot: it needs to be precise. If the key is too wide, the stretcher can split; if too small, the key will fall out (Kuschan 1999). All agree that at least the lower edge of the key should be roughened to prevent the key from sliding too easily out of its slot; some simply leave the key edges rough from the original saw cut. Wingrill makes keys with the long edge serrated to seat it better inside the key mortise (Wingrill 1999). Many conservation stretcher makers also recommend rounding or otherwise reducing the sharpness of the keys' corners to reduce the danger of poking holes in the canvas if the key falls between the canvas and stretcher.

Placement of Key in Stretcher

There is some disagreement as to the angle at which the key should be inserted into the stretcher. If the key has been cut with the grain parallel to the shorter of its two long sides, as is usual, the majority of those asked say the key should be inserted into the opening sitting on its longest edge. This requires that the hammer be held at a slight angle to tap the key into place, also recommended to avoid splitting the key. However, some recommend inserting the key sitting on the shorter of its two long edges. Thus, the short edge to be tapped by the hammer is vertically oriented. This enables the hammer to be slid along the inside stretcher edge to tap the key into place, allowing more control of the hammer and reducing the chance that it will hit the canvas prior to hitting the key. All recommend inserting a thin, light protective sheet behind the key, between the fabric and inside stretcher face prior to tapping with the hammer. This can be a piece of cardboard, ragboard, plexiglass, metal plate, etc. This protective sheet avoids contact by the hammer with the canvas, which can produce bulges in the canvas or characteristic crackle in the paint on the front. Simon Liu recommended using a spreader clamp to insert the key in the slot rather than hammering it in, thereby avoiding the problems of vibration and potential damage to the painting (Liu 1999). Techniques for preventing the key's slipping out are covered in another entry (see section IV.K).

Sarah L. Fisher

Submitted October 2006

Mechanics of a Painting

Review of Research to Date

Introduction

The rheology of a painting is the study of how the materials of a painting flow and deform over time. Most materials are affected by moisture and temperature as well as applied stress. While they may appear static to the casual observer, stretched paintings in particular are conglomerations of different materials constantly on the move. Of course, these movements are generally infinitesimal and hard to observe, but they can add up to very obvious alterations that can be detrimental to the aesthetic intent of a work of art. The most obvious example in a painting is cracking of the paint layer. Therefore, the goal of the conservator is to keep such potentially damaging movement to a minimum.

The amount any material moves is affected by many factors, including, in the case of a stretched painting, applied stress. When a painting on a fabric support is restretched onto a secondary support, considerable force is involved in the stretching process. How is the applied stress of stretching likely to affect the painting? If a stretched painting is maintained taut and in plane, the materials of the painting are under restraint. How does this affect their mechanical response to the inevitable fluctuations in temperature and humidity? Complicating this situation, the stress across the plane of a stretched canvas is not evenly distributed. Although the goal in stretching or restretching is to create even tension, with equal stress along all four sides, this is very difficult to achieve. Moreover, in a conventional rectangular format there will always be a concentration of stress at the corners, as well as other variations in stress distribution. Finally, how the fabric support is attached to the secondary support will also affect the distribution of stress in the painting. Are staples or tacks preferable? Should they be applied in a line or staggered?

However, even fabric paintings attached to a solid, stable support can move as a result of changes in environmental conditions. Often the rheology of a painting is determined as much by the paint or ground layers as the support. The situation is further complicated by the composite and varied nature of the pictorial layers. The conservator of a painting on fabric might be faced with the delicate wash of a Frankenthaler on unprimed cotton or the thick encrusted impasto of a Schnabel. Across the surface of a single painting there might be a thinly painted blue sky and a foreground of heavily impasted grass and flowers.

Different materials respond differently to environmental changes, creating stresses within the layer structure. For example, one must consider modulus and dimensional response, stiffness, glass transition temperature (Tg), age, and time of applied stress for each material; many of these mechanical properties are themselves affected by environmental conditions.

The history and condition of a painting may also affect its rheology. What if the painting is covered with age cracks and the paint layer is no longer continuous? What if the painting is lined with a stiff sailcloth lining? A thick glue size layer may make a painting particularly vulnerable to changes in relative humidity, and a subsequent wax lining may alter that sensitivity. Cracks in the paint layer may affect the layer in the composite structure that is actually carrying the stress in the stretched system. Some examples of aged paintings display perfect preservation where the paint has been firmly supported. Others show damage only where restraint has occurred.

All of these factors should be considered when undertaking a conservation treatment. With this infinite variety, it might seem that determining the rheology of a painting as a whole would be impossibly complex. However, some general understanding of the behavior of the materials involved can help in assessing the effects of many conservation treatments, including stretching, restretching, or lining a painting. It can also help in predicting the response of the painting to fluctuations in environmental conditions.

Overall Tension

Consideration of the mechanics of a stretched painting has led repeatedly to speculation and study on how to reduce the ill effects of such a system. Two major scenarios need to be considered. First, with a painting under stress, elongation of the materials, or creep, may occur over time, resulting in a decrease in tension or slackness of the painting. For example, studies by Berger and Russell confirmed that deformation of canvas under stress is not fully recovered, but plastic deformation is reduced by priming (Berger and Russell 1984). Second, cycling of temperature and relative humidity, a common occurrence with seasonal and even diurnal changes, will result in alternate periods of tension and slackness as the materials of the painting expand and contract with changing environmental conditions. This may by itself result in plastic deformation. Or, in a reverse of the phenomenon of compression set described by Buck, a painting may expand over time because it is restrained only in times of contraction (Berger and Russell 1986, 49–50; Berger and Russell 1994, 79; Buck 1972). In contrast to Buck, where repeated cycling of relative humidity (RH) caused a cumulative compression of the wooden support, Berger found that there was a cumulative elongation of the fabric with repeated environmental change (Berger and Russell 1994, 79). Consideration of both these issues leads to questions of how important it is to maintain a constant tension, what degree of tension is best for a painting, and what is the best way to maintain this constant stress level.

Although long considered a primary goal in painting conservation, the importance of maintaining a strictly controlled environment for the preservation of works of art has been reconsidered by some (Erhardt, Mecklenburg, Tumosa and McCormick-Goodhart 1995). However, significant fluctuations over short time periods are certainly agreed by all to be potentially hazardous to many works of art. For aesthetic reasons, as well as to avoid possible dangers such as the support slapping against the stretcher bars, maintaining a stretched painting in plane has also been a goal. The degree of tension that may be best for a painting is questionable, however. The least invasive way to maintain a desired level of stress is still to maintain a constant environment. As an alternative, Berger, Buck and others have sought to achieve a constant stress level by attaching a painting to a stretcher designed to maintain constant tension (see sections III.B.3, III.B.4.a, and III.B.4.b). Gustav Berger began to look into the relationship between fabric supports and changes in temperature and humidity after working with cycloramas, large paintings that are free hanging and therefore under constant tension (Berger 1980). He noticed that although the dimensional changes that cycloramas experienced with changes in temperature and relative humidity were very large, the condition of the paint on these vast paintings was generally very good. This led him to speculate on the effect of stretching on a painting's response to environmental changes. Using an apparatus he developed to test stress biaxially (in two directions) on a sample of canvas, he found that stress varied 200–300% with changing RH. He concluded that there needed to be a newly designed stretcher that would expand and contract with the painting to counter the loss of tension resulting from creep.

He performed a series of tests with the Berger-Russell Stress Tester starting with raw canvas, then adding sizing, priming, and paint, and finally testing an aged sample painting. He concluded that stiffer supports restrain the movement of the paint film and reduce damage. As a consequence, stiffer supports create greater stresses in the painting and, in general, tension and stiffness are proportional. This led him to suggest the use of self-adjusting spring-loaded stretchers to stretch a painting to just below the yield point (what Berger calls the MST: maximum sustainable tension). However, Berger also cautioned against overstretching. If a painting is stretched with a tension above its yield point, plastic deformation will occur, as the painting simply cannot sustain this level of tension. Hedley addressed this concern as well, by demonstrating the role of stiff lining materials in reducing the dimensional response of paintings (Hedley 1981).

Variations Due to Application of Stress/Stress Concentrations of System

Beyond the question of the overall stress in a stretched painting, consideration must be made of the variations that will inevitably exist across the plane of a two-dimensional painting. This question also presumes some method of measuring the stress in a painting. Several different approaches have been taken in this task. Berger, Hedley, Mecklenburg, and others have sought to physically measure the stress either in the individual materials of a painting, combining these results to achieve the stress levels of the composite, or measuring physical models or samples of actual paintings. Another method, developed by Mecklenburg, involves using mathematical modeling to predict such measurements based on measurements from actual samples.

First, variations occur due to the woven nature of the fabric itself. While testing canvas samples on his biaxial stress tester, Berger noted a dramatic difference in the warp and weft directions, with the weft testing 6 times stiffer than the warp. The behavior of cotton in particular was found to be anisotropic. Second, in a rectangle of fabric stretched on a frame, variations occur not only in the level of stress but also in the kind of stress: tension or compression. In 1975, Hedley published a simple but very interesting study of the distribution of stress across a stretched painting (Hedley 1975). After drawing a grid on a piece of unstretched fabric, the fabric was then stretched and the alteration in the lines measured and used to determine the distribution of stress. He found three very different patterns. The stress in both warp and weft through the middle of the fabric was consistently tension, with an increase of stress at the edges. Along the perimeter of the fabric, in the lines at top and bottom that ran parallel to the weft direction, the stress was generally compression. In a few places, where there was no applied stress in the warp direction, no stress was measured. If the measurement corresponded to the edge of a tack along the warp edge, there was a small area of tension. In the lines running parallel to the sides of the stretched fabric, which ran in the warp direction, there was an even more complex distribution of stress. In the lower third, there was an area of tension, in the middle an area of compression, and in roughly the final third, another area of tension. The areas of compression are all due to the Poisson ratio effect. When a material is put under tension, or pulled apart in one direction, a force is also generated perpendicular to the line of tension. The Poisson ratio relates these two forces. The greatest stress concentrations were measured at the corners in a direction running 45 degrees from the warp/weft directions. This finding is consistent with the pattern of cracking that is often found running across the corners of a painting. Overall the study demonstrated the complexity of the distribution of stress in a stretched canvas. Hedley found a number of factors played a role in the distribution of stress, including the properties of the material itself, the manner in which it's stretched, and the dimensions of the stretcher (Hedley 1975, 25). Complicating the picture is the very heterogeneous nature of the applied strain, in a series of point loads. Mecklenburg used mathematical modeling to provide a more sophisticated picture of these variations in stress, determining variations in the corners vs. the center and even along the edges (Mecklenburg and Tumosa 1991, 182). There is not complete agreement about the source of these stress concentrations, however. Michalski, for example, attributes radial corner cracks to the bias introduced by shrinking in the stretcher bars (Michalski 1991, 240).

As Hedley noted in his 1975 study, the method of attachment of fabric to stretcher will also affect distribution of stress. This factor has been studied extensively by Young and Hibberd. Their 2000 study examined the role of the method and manner of attachment on the stress/ strain distribution within a stretched fabric. They performed two different tests. In one test, small stretched canvases were stressed biaxially; in the second, strips of canvas were tested uniaxially. A contour map of the strain distribution was made using electronic speckle pattern interferometry (ESPI). Their studies reached several interesting conclusions. The first concerned the corners of a stretched painting, long considered to be a place of concentrated stress and shown by Mecklenburg and others to result in distinctive diagonal cracking in paint and ground layers. They noted that this pattern of deformation was found on paintings that had never been keyed out and could therefore be the result of fluctuations in environmental conditions alone. They speculated that the considerably lower strains at the corners of their test paintings were likely the result of the tight “hospital corners” used in folding the fabric at the stretcher corners. They also studied the effect of lacing the fabric to a stretcher and found this method resulted in no concentration of stress at the corners.

In a comparison of the relative merits of staples vs. tacks, they found less concentration of stress with tacks. Tearing was also a problem with staples, particularly if the staples were not absolutely parallel to the edge. If they were not perpendicular to the direction of the load, stress would be concentrated in one of the staple “legs” and tearing was much more likely. Similarly, up to a certain load, the staple would act as a “bridge,” meaning the load would be distributed along the entire length of the staple. However, above a certain load, slippage of the fabric under the staple's middle would occur, and stress would be concentrated at the two legs, which would often result in tearing. The sharp rectilinear edges of the staples certainly play a role in this; therefore rounded “wire” staples are preferable. Tearing was not a problem with tacks. Historically, failure of the fabric around the tack was the result of deterioration caused by the rusting of the tack, a problem easily eliminated with the non-ferrous tacks available today. Closer spacing of either staples or tacks led to reduced stress/strain concentration. Staples applied diagonally to the tacking edge showed no difference in stress distribution to staples applied parallel to the edge. Staggering the staples added stress concentrations, however, probably due to the greater amount of fabric in front of some staples.

In comparing attaching the fabric on the face, on the side, and along the back edge, the further the attachments were from the picture plane, the more even the distribution of stress. There are probably two factors at work here: more fabric to even out the stress and the friction of the stretcher bar. In general, the authors calculated the side and rear attachment resulted in 67% and 45% of the stress of front attachment, respectively. The application of card between staple and fabric was also studied. A continuous piece of card, running the length of the edge, was most effective at distributing stress more evenly, and thicker card was more effective than thin card or fabric tape. Probably the stiff card works like the “bridge” of the staple, distributing the stress over a larger area and mechanically resisting cusping. All the canvasses attached using staples showed visible cusping, and the authors found secondary cusping could be formed by restretching as little as four weeks after priming. Preliminary indications show that care must be taken in restretching to minimize the potential of creating secondary cusping. Further research on this topic is currently being carried out by the authors (Young and Hibberd 2000, 212–220).

The Composite Nature of a Painting

In addition to the differences in stress across the plane of the painting, variations occur within the thickness of the painting composite, due to the range of application and to the different ways that the layers—fabric, size, paint, varnish, and wood—respond to changes in temperature and humidity and exhibit differing degrees of stiffness or resistance to deformation. First, one must determine as nearly as possible the likely behavior of each material. Second, while some limited testing of actual paintings may be carried out, in general this is not possible. One must figure a way to weight and add these behaviors to predict the behavior of the whole. Superposition, mathematical modeling, and physical models are all ways to estimate behavior. Different factors determine which layer in each painting is the one that carries the stress for the composite. The strongest or stiffest or thickest or most continuous layer might be the one that dominates the behavior of the whole and determines the behavior of the painting. Finally, the way that the layers interact, with their varying tendencies to expand or contract, and the influence they exert on each other will also help determine the behavior of the composite.

In 1982, Mecklenburg put together a rather simple but groundbreaking study of the effect of changing temperature and relative humidity on the basic materials commonly found in traditional oil paintings: fabric, glue, and oil paint. Mecklenburg assessed the effects of environmental change on the mechanical properties of each material. He also considered the dimensional response of each material to added stress. Estimating the response of a material to changes in temperature (T) and relative humidity (RH) in a system under stress can be complicated. Several different factors are involved in assessing the response of the various materials to a changing environment: mechanical properties such as strength, stiffness, modulus, plastic vs. elastic behavior, dimensional response or thermal and moisture coefficients, and time. Changes in both temperature and relative humidity can create stresses in the materials of a painting that can then lead to plastic deformations.

Examining the effect of changing temperature, Mecklenburg found that both the modulus and the thermal coefficient of expansion affect the level of stress in a cooling material. In assessing the effects of variations in RH, Mecklenburg considered two related mechanical properties: stiffness and strength. Dimensional responses to changes in temperature, though significant, were all much smaller than responses to changing RH (Mecklenburg and Tumosa 1991, 181).

In addition to dimensional response, the modulus of each material must be considered in order to determine the level of stress induced by loss of heat or moisture. The modulus is, in turn, a function of both the T and RH. If the thermal coefficient of expansion and the equilibrium modulus are known, the level of stress can be predicted (Mecklenburg and Tumosa 1991, 178). Mecklenburg found the behavior of most artists' materials to be elastic. Hedley was particularly interested in the balance between elastic and plastic behavior in materials exposed to changing RH. He found that moisture absorption lowered the glass-transition temperature (Tg) of paint films, making it possible to exploit their plastic behavior to treat deformations.

In contrast, low moisture absorption led to more elastic, brittle behavior. The higher the oil medium content in a paint film, the more able it was to absorb moisture (Hedley 1993).

Working from the assumption that the size, ground, and paint carry most of the tension in a stretched painting and recognizing that all these materials are amorphous or semi-amorphous polymers, Michalski utilizes the abundant scientific information on the properties of commercial paints and polymers gathered from the paint industry. The notes from his article in Art in Transit are a good source for suggested reading in industrial paint literature (Michalski 1991). He found the unifying concept for all of these materials to be the glassy/rubbery transition (Michalski 1991, 223). Low temperature, low humidity, high pigmentation, and aging all increase the stiffness of these materials and the likelihood that they will crack, particularly when exposed to short-term loads. He is also concerned with the tension resulting from the tendency of these materials to shrink with low temperature or humidity.

Further research specifically on the mechanical properties of modern paints has been performed by Hagen and others. They studied the effects of age, T, RH, and leaching on cast films of poly(methyl methacrylate-co-ethyl acrylate), obtaining stress/strain curves under controlled conditions. Decreasing T and RH both increased stiffness and strength. Decreasing T had the greatest effect, resulting in embrittlement of the paint film (Hagen et al. 2007).

Mechanical properties such as Tg are affected by temperature, humidity, applied stress, and age. Time of applied stress and temperature are connected—high strain can occur with low temperature or fast stress. The relationship between plastic and elastic behavior is important in determining response. If glassy, there is about 3% elongation before either brittle (crack) or ductile (flow and crack) response (Michalski 1991, 224). At higher humidities, water molecules can act like a plasticizer.

Mecklenburg, Michalski, and Hedley all stress the importance of time as a factor. Time and temperature together influence glassy/rubbery behavior. Michalski also investigates painting response times. The thermal response half time (or the time required for half of full response) of low impasto and ground is 10 minutes, while for wood panels it is two hours (Michalski 1991, 232). If the heating is caused by direct sun, times will be shorter and gradients steeper. The humidity half times for linen, size, and gesso range from minutes to hours; for oil grounds and paints, from hours to days; and for wood, from days to weeks (Michalski 1991, 232). There is also a differential between the front and back of the painting, which can be modified by the use of a backing board.

With the combination of studies by conservation scientists and technical information from the paint industry, a broad base of information is developing.

Fabric

In 1980, Hedley published an article on artists' canvases that discusses the mechanical properties of cotton and linen and the problems these pose for the stability of stretched paintings.

Except at very low RH or RH above 80%, the fabric support generally does not determine the physical behavior of a painting.

Material studies by Mecklenburg in 1982, confirmed by Hedley and others, indicate that the modulus of the fabric is only significant at very high RH when other materials of a painting are soft and gel-like. He found that fabric has low stiffness at low humidity, while stretched fabric is very stiff at high RH. Fabric tends to shrink below 80–85% through drying; above this point it also shrinks, but here due to alteration of the crimp. Hedley gives a detailed description of the mechanism of shrinkage and the importance of canvas geometry in predicting behavior, in particular the tightness with which each thread is wound as well as the tightness of the weave (Hedley 1993, 112–114). Shrinkage can also occur with loss of heat, but in general shrinkage due to lower temperatures is balanced by increased moisture content.

Berger, Mecklenburg, and Hedley all note that the first time a fabric is exposed to high humidity it responds differently from every subsequent exposure. Berger went so far as to suggest that if a fabric was not sufficiently wetted before stretching, it could not provide adequate support for a painting for more than 20 or 30 years.

Sizing/Gesso

The sizing/gesso layers can play a dominant role in the physical behavior of many paintings. Early studies by Berger found that priming increased the stiffness of fabric, but only if the priming remained uncracked. Berger also discussed the dominant role of the glue sizing in younger paintings when the paint is still relatively soft and the glue layer, if it is a reasonably heavy one, has not yet cracked and become discontinuous. Too concentrated a glue layer could result in severe cracking of the painting, and too weak a solution of glue could create a “shrinker,” a painting that could react violently to exposure to moisture. This is probably due to the failure of the size to permeate the fabric with a sufficient layer to reduce its moisture response. Hedley showed high RH reactivation of a stress relaxed glue layer will at least triple its stress response to RH when restrained in a painting over 100 years old (Mecklenburg and Tumosa 1991, 187).

Mecklenburg found that glue exhibited quasi-plastic behavior, as its yield stress was easily reached at low RH, after which it would not sustain any stress. However, if a high RH was reached, the glue could reactivate itself and potentially cause significant damage to the painting. He also found that glue shrinkage on drying is a major source of stress in paintings exposed to changes in RH. This dimensional response has two ranges of behavior, one from 85% to 70% where the rate of shrinkage is more rapid, and one from 70% –0% where the rate of shrinkage is less (Mecklenburg and Tumosa 1991, 175). The dimensional response of gesso is similar, but the different ranges occur above and below 80%, and swelling is affected by the ratio of chalk to glue. Michalski determined that below 70% RH, glue shrinkage is five times greater than the average oil paint. Films shrink less in length than in thickness (Michalski 1991, 226). For typical gesso recipes, shrinkage is several times less than oil paint (Michalski 1991, 226). Michalski found glue and gesso to have complicated behavior with stress strain curves exhibiting two-stage elastic behavior. Both of these phases are glassy to shock and vibration at moderate RH (Michalski 1991, 230). Exposure to high humidity can cause glue to age rapidly through the formation of crystallites (Michalski 1991, 230). Michalski details three separate mechanisms that can increase the tension in a glue film after high RH:

1. If prevented from swelling, the glue may plastically compress, then try to shrink on return to moderate RH.
2. The glue may physically age, resulting in compaction of the network and formation of crystallites.
3. The glue may previously have been stretched, creep occurred, then high RH may revive tension, plus cause swelling.

Gesso has similar stiffness to glue. The ground layer is particularly vulnerable because of over-pigmentation and has the lowest elongation at break. In addition, because it is hidden below the surface of the painting, damage can take years to become visible, as cracks are propagated in the paint layer. In any case, once a painting is sized, the glue, not the fabric, determines the support's dimensional response (Mecklenburg and Tumosa 1991, 175).

Paint

Like Mecklenburg and Michalski, Berger found that paint films respond in a rather dramatic way to relatively small changes in temperature. For example, in one study he found a larger stress was generated by a 3 degree C change in temperature vs. a 38% change in RH (Berger and Russell 1994, 76). Mecklenburg's computer modeling led him to conclude that low temperature is more likely to crack an undamaged painting than low RH. Similar results were found in a recent study of modern acrylic emulsion paint (Hagen et al. 2007). Michalski found that a range of materials including solvents, water, and naturally occurring by-products of paint degradation can affect behavior by acting as plasticizers. Pigment Volume Concentration (PVC), or the percent of pigment per volume of paint, is a major characteristic affecting the stiffness of a paint film. PVC above 30% makes much more brittle paint. Artists' paints are 30–60% PVC (if “buttery”).

In paint industry assessments of various paints, two different measures are used to determine the point of failure of a given paint film: strength and elongation at break. Stiff materials are stronger but more brittle. Less stiff materials will deform at lower stresses. Both properties must be considered in assessing the response of a painting to applied stress. According to Mecklenburg, the paint film stiffens and strengthens with age. The dimensional response, or moisture coefficient, decreases with an increase in modulus, making aged paint less dimensionally responsive to changing RH. He found that oil paint had two distinct ranges of behavior: from 0%–70% there is little swelling; from 70%–95% swelling increases (Mecklenburg and Tumosa 1991, 174). In general, stiffer paints showed less dimensional response than more flexible ones.

The effect of RH on the stiffness of paint seems fairly uniform. All paints double their stiffness when dried from 70% RH to 0% RH. Above 70% RH, their behavior varies according to the pigment involved. The glass rubber transition for oil paint begins near −30°C and ends before 0°C for linseed, walnut, and poppy oil, with or without pigments, with or without aging (Michalski 1991, 228). For acrylic emulsion paints, the most common copolymer used is Rhoplex AC-33, which is 40% methyl methacrylate, 60% ethyl acrylate. It forms a tough and resilient film but becomes glassy near 5°C.

When an acrylic emulsion painting dries, the volume of water lost equals the volume of shrinkage. Paint and glue are generally isotropic, so shrinkage occurs equally in three directions, with one-third of the total in each direction. By 70% RH, the average oil paint has adsorbed one molecule of water for each molecule of oil (Michalski 1991, Note 21). This is an average; Mecklenburg's studies show some paints with triple this value and others with only half. The type of pigment affects the absorption behavior of the oil. Above 70% RH, “moisture clusters in the oil medium and builds up on the pigment surface, so swelling is much greater” (Michalski 1991, Note 24). This behavior is far from uniform, however, as paints can behave differently on different cycles or not be isotropic.

Varnish

With a Tg near 50°C, natural resin varnishes are the glassiest of all painting materials. Luckily, they rarely carry the stress of the painting. However, cracks in the varnish can propagate cracks in the paint layer. Luckily, when they are not pigmented, they can experience elongation at break upwards of 3% (Michalski 1991, 231).

Wood

The rheology of wood, a complex topic, will be covered more fully in another chapter. Mecklenburg considers the response of wood, both as panels and as used in wooden stretchers. Here, it is important to identify both the type of wood and the grain, as wood can respond on average .3% in the longitudinal, 3.6% in the radial, and 5.9% in the tangential direction (Mecklenburg and Tumosa 1991, 176).

Whole

Most researchers have sought to understand the rheology of the complex composite of a painting by studying the individual components and summing their behaviors. Mecklenburg's work on the mechanical properties of materials in the early 1980s involved the testing of each material individually, using fabric samples and cast films. Hedley looked at the superposition of Mecklenburg's stress/strain curves and compared this with the results of tests on five naturally aged painting fragments. He was only able to test them uniaxially, but the results compared favorably. He also continued studies using cast films.

Beginning with more elaborate and extensive research into the mechanical behavior of each layer of the painting composite, Mecklenburg has combined this research with mathematical models designed to predict the behavior of the entire painting in response to environmental change. A general summary of this research can be found in several places (Mecklenburg 1991, Mecklenburg et al. 1994). Modeling of RH and T effects was first attempted by Mecklenburg (Michalski 1991, Note103). Although the model was based on a simple layer structure, the effect matched closely with observed behavior. Michalski offers a few general equations for determining the stress in a painting:

1. Painting tension = sum of tensions in each layer
2. Each layer tension = sum of tensions from all previous events
3. Tension from each event = shrinkage or stretch x stiffness for that RH, T, and time interval

Mecklenburg has found some interesting anomalies, however. It appears that when a painting is cooled, the materials react independently of one another, while there is considerable interaction in a painting undergoing desiccation.

Although Richard Buck was concerned for much of his career with the behavior and treatment of paintings on wood panel, in discussing the warping of painted panels, he brought attention to the idea that the outer “skin” of a painting, when exposed to environmental changes, would respond differently from the inner core of the panel, and even relatively small changes in environmental conditions, recurring repeatedly, could create significant plastic deformations in the painting. Several studies by Berger focused on the composite behavior of paintings, both between layers and within the layer of the paint, and the analogy in behavior to a solid's surface “skin” and its inner core. For example, the outer skin, exposed to increased moisture, attempts to swell but is constrained by the inner core. When the humidity decreases again, the skin, which has been plastically deformed, is not as long as the inner core, and tension develops. In testing with computer modeling, Mecklenburg found that with changing T, the layers of a painting did not respond equally and restrained each other. A stiff fabric support can reduce interlayer stresses.

History/Condition

History and condition can also have a profound effect on the behavior of a painting in response to changes in temperature and humidity or applied stress. Berger warned that, “the physical characteristics of canvas and paint films are transformed by every treatment and every exposure to stress” (Berger and Russell 1988, 194–5).

Cycles of high and low humidity can create cracks in the paint and ground layers. Patterns of cracking can indicate patterns of tension in the painting. In general, Y junctions occur at the beginning of cracks, and Ts at the ends. Different layers can have the same pattern with varying spacing. There are often fewer cracks over the stretcher bars and deep cracks along their inside edges.

Initial aging of the paint results in cross linking of the polymer film. Aging also degrades the “microstructure” of the paint film: ultraviolet light or moisture can damage the surface or break pigment-medium bonds, weakening the paint. “Physical aging” can also occur as the randomly arranged molecules settle into a more closely packed system, creating a stiffer, glassier film (Michalski 1991, 225). Solvent exposure can result in the removal of plasticizers, rendering the paint film more brittle (Erhardt and Tsang 1990, Hedley et al. 1990). Existing cracks in paintings on canvas provide potential concentrations of stress, complicating the painting's response to additional stress. Paintings with existing cracks are much more susceptible to damage with fluctuations in RH. Berger showed that cycling of RH can result in cupping and that, “large stress changes can result from small cyclical changes” in T and RH (Russell and Berger 1982, 7). Mecklenburg found that with fluctuations in RH from 70% to 50% at 23°C, the stress concentrations at the cracks easily exceeded the paint's strength (Mecklenburg and Tumosa 1991, 189). Berger found problems particularly arose when only one phase of deformation (i.e. compression or elongation) occurred, as a result of either location or geometry. For example, if there is a crack in the paint film, the affected paint can undergo compression, but not tension, as the film is not continuous. He referred to this point of concentrated stress as a “deformation valve.” These anomalies, combined with cycling RH changes, result in defects such as cupping and blistering. Mecklenburg's computer modeling concurred that cracked paintings can tolerate less stress. Stress concentrations occur at cracks. Once a crack starts, much less stress is then required to make it grow.

Lining, by attaching a new, strong support to the reverse, can alter where the stress is carried through the painting composite and introduce greater stiffness. High stresses from glue contraction at low RH can create deformations that a stiff support can reduce while stiff linings can protect a painting from environmental-related damage. Mecklenburg concluded that stiff linings attached with adhesives unaffected by RH will reduce the formation of new cracks caused by changes in RH. These allow tension to develop without deformation, but do not limit stress. The lining adhesive can also have an effect on the behavior of a painting. Mecklenburg found glue linings make paintings more susceptible to damage in a changing environment. In 1975, Hedley presented a study of the effect of wax-resin impregnation on the tensile properties of fabric.

Lining with a stiff fabric can also reduce the degree of tensioning required to achieve a desired appearance of tautness. On the other hand, stretching, even minimal stretching (such as keying out), lowers a painting's tolerance of environmental changes. “The expansion of paintings by any corner device, even seemingly small amounts, renders a painting more susceptible to damage due to either cooling or desiccation”(Mecklenburg and Tumosa 1991, 189).

Many of these connections can help to inform future conservation treatments. “Heat, moisture, tension and deformation are the main causes of decay in canvas paintings. They are also the most important tools in their treatment” (Russell and Berger 1982, 8).

Jane Tillinghast Sherman Submitted January 2007

Practical Pros and Cons of Keys vs. Other Joint-Adjusting Mechanism

Keyed Stretcher Designs

Keyed stretchers are usually joined by a mortise and tenon with slots to accommodate keys (see section III.B.1). As the keys are forced into these slots, they exert force against the members, thus increasing the dimensions of the stretcher. The tension created in the fabric is primarily located at the corners of the stretched fabric and at any keyed cross-members. The choice of a keyed stretcher may mean that the original stretcher (possibly with modifications) may be used.

Pros

• Keyed stretcher members are readily available in a number of precut sizes and thicknesses.

• Custom-made sizes are available from some manufacturers.

• The basic construction of a keyable mortise and tenon stretcher is relatively straightforward, often allowing an original stretcher to be disassembled and repaired fairly easily.

• New keys are easy to cut and replace.

Cons

• Some commercial members are cheaply made or of inferior wood. Precut sizes may not be of the exact dimensions needed for a particular painting.

• The construction makes the joint vulnerable to splitting due to the insertion of keys too far into the joint or the use of keys that are too thick in relation to the size of the slot.

• Keys may break in the joint and be difficult to remove.

• If the relative density of the wood used for the keys is very different from that used for the stretcher members, as the keys are driven in, the less dense wood may compress rather than exert the expected force on the other wood.

• As the wood of the stretcher shrinks with age or low humidity, keys may fall out, often becoming lodged between the fabric and the stretcher member. (The keys should be held in place by some method that prevents them from falling out but allows for future adjustments.)

• The physical force of the blows required to drive the key in and expand the stretcher may damage paintings. (The fabric must be protected when driving in keys.)

While keys may be driven in to further expand the stretcher, there is no simple method for reducing the dimensions of the stretcher.

Expansion Bolt Stretchers

The design of the stretcher is similar to the traditional mitered joint, keyed stretchers. In place of keys, an expansion bolt is set in each corner perpendicular (at a 45° angle) to the miter. Turning the bolt controls the contraction and expansion of the stretcher. The tension created in the fabric is primarily located at the corners of the stretched fabric and at any cross-members that include an expansion bolt. Use of this stretcher may require replacement of the original, although some contemporary artists are now using this stretcher (see section III.B.2).

Pros

• Stretchers are custom ordered to desired dimensions.

• The stretcher adjustment is done on the back face of the stretcher; therefore, damage to the fabric is less likely.

• Adjustment involves turning the bolt, an action that could be more controlled than the tapping in of keys, as in a keyed stretcher.

• Reducing fabric tension is accomplished by reversing the bolt.

• The amount of expansion is more easily observed, i.e., each bolt turned one and a half turns, etc.

This type of stretcher may more easily accommodate paintings that are out of square because opening of the corner joints can be more controlled.

Cons

• The stretchers are not readily available commercially.

• In order to turn the key, a special rod must be used; however, other properly sized tools, such as an awl, may be used. (This may also be considered an advantage, in that it reduces the chance of indiscriminate tension adjustments.)

• The turning edge may be sharp, requiring modification (i.e., sanding) before stretching.

• An open joint is created at the corner when expanded, which may reduce the overall structural stability.

• The expansion occurs in both directions at the corner, as it does in a traditionally keyed stretcher; this may not be desirous if the fabric tension only needs to be adjusted in one direction.

Spring Tension Designs

The most common form of spring tension stretcher is the ICA stretcher, introduced by Richard Buck in 1950, and available until 2001 under the name Superior Spring-Stretcher (they are no longer manufactured) (see section III.B.3). These stretchers are built with plates and springs in the corners that exert continuous tension on the fabric as it responds to environmental fluctuations. The tension created in the fabric is primarily located at the corners of the stretched fabric. Use of this stretcher requires replacement of the original, although some contemporary artists are now using this stretcher in their work.

Pros

• Stretchers are custom ordered to exact dimensions.

• The plates in the corners increase the rigidity of the stretcher.

• The tension may be increased or decreased by adjusting the tension on the springs.

• The spring system maintains even tension on the fabric in changing environmental conditions; this may be particularly useful in situations in which the climate cannot be perfectly controlled. However, tight framing or the use of a backing board attached to the stretcher can restrict stretcher movement and limit this benefit.

• As with expansion bolt stretchers, the amount of expansion is easily observed as adjustments are made to the springs.

Cons

• Not readily available commercially, but may be custom ordered.

• Adjustment of the springs requires a specially sized tool, which may provide an advantage in limiting inappropriate adjustments.

• The continuous tension of the stretcher can damage weak fabrics. (These stretchers should only be used on lined paintings or on new paintings executed on strong fabrics.)

Overall Bar Adjustment Stretcher Designs

This stretcher category includes the Individual Bar Control (IBC) Stretcher developed by Stan Phillips in the 1990s, and the Self-Adjusting Continuous Tension (CT) Stretcher developed by Gustav Berger in the 1980s (see section III.B.4b). The IBC stretcher consists of a fixed inner frame that is permanently assembled with a mortise and tenon corner held with screws. On each side, an outer movable bar with a profile similar to that of a traditional beaded stretcher is held with wooden dowels (an early design used fiberglass dowels). The stretcher may be expanded by adjusting bolts located at intervals along the length of the bar between the fixed inner frame and the expanding outer frame.

Berger's CT stretcher consists of a fixed, strainer-like frame. Only the upper stretcher member includes a separate, movable bar, and tension in this bar is adjusted by means of springs. The painting is affixed to the stretcher only along the top and bottom edges. The vertical edges of the fabric have pocket strip linings attached to them. A rod is inserted in each pocket, which holds the vertical fabric edges straight. Horizontal fabric tension is achieved by springs between the rods and the stretcher frame.

In both stretcher designs, the fabric is expanded at all points along the perimeter, and expansion is not focused at the corners. Therefore, while the amount of tension present in the canvas may vary at each point of adjustment, it is not concentrated solely at the corners of the stretcher. It should be noted that the IBC stretcher works by hand expansion of the stretcher members and does not provide continuous tension, while Berger's CT stretcher provides continuous tension to the stretched painting by means of springs.

Pros

• Exact sizes and custom shapes may be crafted to accommodate the irregularities of a painting.

• The ability to control stretcher expansion along the length of the stretcher also provides adjustments for irregularly shaped (sized) fabrics.

• In the CT stretcher, the fabric is allowed to move in response to environmental fluctuations, like the spring tension stretchers.

• The tension of stretching is distributed evenly around the perimeter of the painting, providing more uniform overall support for the fabric.

Cons

• The IBC stretcher bars cannot be expanded a great deal. Excessive expansion may cause the outer moveable bar (stretcher frame) to torque or bow.

• For exceptionally large IBC stretchers or for strong fabrics, the distance between expansion bolts must be considered. Too great a distance between bolts may cause bowing of the outer moveable bar (stretcher) between expansion bolts.

• A specially sized tool is required for bolt adjustments in the IBC stretcher and spring adjustments in the CT stretcher.

• Use of either stretcher requires replacement of the original, and not many contemporary artists are using them at this time.

• The CT stretcher requires the addition of pocketed strip linings to the vertical edges of the painting to hold the vertical rods.

• Successful use of the CT stretcher requires careful adherence to a specific method of mounting.

• The IBC stretcher is heavier than most stretchers due to the inner frame construction. The CT design is in limited production, primarily through Laszlo Cser.

Stan Phillips is no longer producing the IBC stretcher, but a version is being produced by Chassitech of France.

Nancy Pollak

Submitted November 2006

Keying Out vs. Restretching

Out-of-plane distortions caused by uneven tension can sometimes be addressed by expanding the corner joints of the secondary support. In some circumstances restretching, a more invasive and complicated procedure, is necessary to address improper tensioning. Although the two procedures may each be utilized, they are not interchangeable—each addresses specific problems.

The basic cause of the planar distortions must first be evaluated. If the picture is lined, it is important to determine the security and strength of the bond with the lining fabric. The condition of the tacking margin turnover and whether it is part of the original or a lining fabric must be carefully assessed. The degree of planar distortion to be corrected is a factor, keeping in mind the overall size of the picture (a larger picture will require greater dimensional change to return it to planarity). The age and flexibility of the primary support is a key element. The condition and thickness of the paint and the presence or absence of stress cracks in the paint layer are also important factors.

Keying Out

In general, keying out a painting is a less invasive and less complicated procedure that requires less time than restretching. If the secondary support is keyable, this is often an appropriate choice for unlined paintings. Keying out is especially desirable for unlined pictures because it retains the original support materials as employed by the artist. However, keying out does nothing to equalize stresses on unevenly stretched fabrics. Another disadvantage is that the impact is greatest at the corners, and if the distortion to be corrected is large or the picture has previously been keyed out, it can open the corner miters too much, leaving them unstable.

If the slackness or bulge has been caused by fluctuations in humidity, the exhibition conditions should be assessed in connection with the planned treatment. If the picture will be returned to an unstable climate, it can create great stresses on an older fabric, even if the keying out has been completed with care.

Restretching

Often part of a complex program of treatment that requires removal of the painting from its stretcher, restretching can correct fundamental problems in the painting structure. It is a more intrusive, complicated, and time-consuming technique than keying out, but may be employed to correct larger or more pronounced distortions. It may be necessary, for proper tensioning of the composite structure of lined paintings, to replace an inadequate strainer/ stretcher or to correct faulty stretching technique.

Disadvantages of restretching include introducing new tensions or redistributing the tensions in a system that has achieved equilibrium. Since it requires removal from the present stretcher, the process will cause damage to the tacking margin, especially at the already vulnerable fold. Restretching a picture that has not previously been restretched will alter the history of the artist's use (and possibly choice) of materials.

Catherine Metzger

Submitted September 1999

Mounting and Remounting a Canvas to a Stretcher or Strainer

A Review of Current Methods and Materials

To ascertain current attitudes concerning the mounting/remounting of canvas paintings on stretchers, an informal survey was conducted (Ameringer 1999). The results of the questionnaire are discussed below. In the United States, as was also found by a survey of UK conservators (Young and Hibberd 2000), the method of attachment appears to be predominantly a matter of personal and/or institutional preference related to the nature of the treatment

Use of Tacks vs. Staples

In a study of canvas attachments, little difference was found between the strain distribution of tacks and staples on a “macro” scale. Under normal loading the head of the tack, not the shank, produced localized strain concentrations, while the staple acted as a continuous area of restraint across the bridge with no localized strain due to each leg (Young and Hibberd 2000).

The vast majority of conservators surveyed report that they generally replace original tacks with tacks and original staples with staples.

• While the insertion of nails requires more stress to the painting (multiple hammer strikes vs. a single staple gun “shot”), most conservators report using tacks when tacks were present originally or are felt to be more appropriate for the painting.

• To reduce the vibration of driving in a tack, a large brad driver or specially constructed pair of pliers can be used to gently “push in” the tack.

Reuse of Original Tacks and Nails

Most respondents to the 1999 questionnaire (Ameringer) indicated that they would reuse original tacks, nails, etc., only if they are unusual. One respondent reported using original tacks in a select area. Some questions to consider:

• Are the tacks/nails in a condition that they can be safely reused?

• Are the tacks/nails unusual or unique?

• Do the tacks/nails appear to be original?

• Is placement of original tacks/nails sufficient to provide even tension or should they be supplemented by new tacks?

Use of Copper and Aluminum Tacks to Replace Iron or Iron Alloy (Steel)

Generally, iron or steel tacks are replaced with less corrosive and more stable copper or aluminum tacks. Corrosion of the tacks can degrade the canvas and make removal difficult. The majority of conservators surveyed report the use of copper or steel tacks; very few report the use of aluminum tacks.

• Copper and aluminum are more stable and less prone to corrosion.

• Copper is a softer metal and may be more difficult to drive into harder woods.

• Some conservators prefer steel tacks for use with magnetic hammers.

• Tacks can be isolated from the surface of the tacking margin by using paper (blotter or pH-neutral) or Mylar®. These blotters or barriers also facilitate later removal. Twill tape can also be used to isolate tacks. An isolating layer may be particularly appropriate if the tacking margins contain paint or part of the design.

Use of Staples

There are numerous types of staple guns and therefore staple types. Some of the more commonly used ones include Swingline® and Arrow. Several conservators mention the use of Monel® staples, which are noncorrosive steel and more costly.

• Staples can also be isolated by using paper blotters or tape, such as twill, linen, or Tyvek® tape.

• In some cases, while tacks may be used to secure the tacking margin to the stretcher, staples may be employed to attach excess canvas or lining fabric to the reverse of the stretcher.

• Care should be exercised when using staples to insert them correctly. If the staple is not inserted absolutely perpendicular to the loading direction, the strain concentration for staples is found to be higher. This is thought to be due to tearing/stretching of the canvas around the staple legs, causing an uneven load (Young and Hibberd 2000).

• Once the canvas between the staple legs begins to slip, staples become a less effective means of attachment, with high strain concentrations around the legs (Young and Hibberd 2000). It is therefore important that the staple be inserted fully and properly.

Use of Adhesives

Using a means other than tacks or staples is more common when restretching a lined canvas. In such cases, the adhesive used for lining generally determines the type of adhesive used to attach the canvas to the stretcher, i.e., for paintings lined (or edge-lined) using ethylene vinyl acetate (EVA) adhesive. EVA adhesive may be used to secure the painting to the stretcher while for wax-lined paintings, wax is often used. In some cases, an additional means of securing excess canvas to the reverse of the stretcher may involve the use of paper tape, poly(vinyl acetate) (PVA), glue, wax, or EVA adhesive. Some conservators feel that securing canvas to the stretcher with a continuous adhesive film such as an EVA adhesive creates a more even distribution of stress forces. Others feel that the difficulty of removal outweighs the advantages; undoing a tacking edge that has been adhered to the stretcher can be “messy” and makes the retreatment more difficult.

Placement of the Tack/Staple

In a study on canvas attachments, the spacing and location of the attachments significantly affected the strain distribution. The study found that closer fixing gives a more even distribution and that strain irregularities diminish successively with attachments applied to the front, side, and reverse of the stretcher bar. No significant differences were found between diagonally and horizontally placed staples. However, staggered placement of either tacks or staples produced a greater variation in strain.

In the 1999 questionnaire, it was reported that tacks/staples were generally replaced in a similar manner to their original placement. Several conservators surveyed report reusing original tack holes when possible. Reuse of original staple holes appears a less common practice.

• Some conservators feel that replacing tacks/staples following the original placement helps recreate stress forces similar to those originally on the painting. Conversely, other conservators feel that an even, regular placement—in some cases, adding additional tacks/staples—creates a more even distribution of stress forces.

• Tacks/staples are most commonly placed in the tacking margin. In rare cases, tacks may be inserted into the face of the painting as part of the artist's original construction. For example, two paintings by Mantegna (Presentation in the Temple, Berlin, Staatliche Museum, and Christ Mocked, Paris, Louvre) show evidence that the canvas was nailed to the front of the strainer with iron nails. Some early paintings in the National Gallery London also show clear evidence that the canvas was nailed on the front face (Young and Hibberd 2000, 212). A number of 20th century artworks have staples inserted at the reverse of the stretcher.

• Tacks are usually placed in a more-or-less horizontal line, although placing tacks in a zigzag pattern is sometimes seen in keeping with guidelines set forth in an Intermuseum Conservation Association (ICA) Handbook from the early 1980s. As noted above, this placement results in a greater strain variation.

• Staples are generally inserted at an angle so that the staple is diagonally oriented with respect to the canvas weave. This is felt by some conservators to be a more secure placement with less tearing of the canvas.

• Some conservators at the Tate Gallery advocate placement of staples on the reverse of the stretcher as they feel it reduces stretching-related paint crackle and more evenly distributes stress forces (Booth 1989). This is in keeping with the results of a canvas attachment study. The study found that as the points of attachment were moved from the face, to the side, to the reverse of the stretcher, strain irregularities diminish successively as they are redistributed within the canvas. This placement was also found to reduce the load on the attachments because of the friction that occurs at the edges of the stretcher bars.

• The 2000 UK study found that the use of a 2-ply card between canvas and staple gave a more even strain distribution. The card extends the area gripped by the staple. Continuous cards provided the most effective grip. (Gripping is reduced if less rigid card stock or cotton tape is used.) (Young and Hibberd 2000) (see section IV.H).

• Young and Hibberd also found that “the restraint imposed by tight corner folds reduces the high load which would be imposed on attachments near the corners if a loosely folded corner were keyed out” (Young and Hibberd 2000, 219).

Charlotte Seifen Ameringer

Submitted October 2006

¹ For brevity, the term “stretcher” will be used in this submission to denote both stretchers and strainers. ¹ Examination of antique stretcher and strainer corners from the collection of the Paintings Conservation Laboratory of the H. F. duPont Winterthur Museum, Winterthur, Delaware. Winterthur has over one hundred examples of stretcher and strainer designs collected over the years in the form of cut corners. They comprise American, European, and Chinese examples from the 18th century to the present. ² Personal experience of author, Charlotte Ameringer.

Appendix

The following appendix is not a comprehensive list of all woods used to make stretchers and keys. The ones listed below are those that conservators use or encounter most frequently.

Softwoods

Cedar is exceptionally light in weight when seasoned, but the timber is also strong for its weight. The straight grain of the wood makes cedar very stable, and it has a good surface finish (Edlin 1969, 110). Cedar is also resistant to termite and fungus attack (Bramwell 1982, 235).

Douglas fir has a fine texture. It is soft, easy to work, and lightweight (Edlin 1969, 116). It is strong and moderately resistant to decay (Bramwell 1982, 264).

Pine has no pores or vessels in its internal structure, so the wood from a pine tree is even-grained. The timber is light, soft, easy to work, yet strong. Since knots usually occur in the tree in groups, with knot-free timber between each group, it is possible to harvest defect-free timber in long lengths (Edlin 1969, 144).

Redwood, also known as Scots pine, is of medium weight and dries well, resulting in timber that is stable in use. It is a strong wood, although it is not resistant to decay (Bramwell 1982, 264). At the time of research, this wood was used for ICA Spring Stretchers; however, at the time of publication, the ICA Spring Stretcher is no longer in production.

Hardwoods

Basswood is also known as lime. The wood from the linden tree is called basswood in America and lime in Europe. In this wood the pores are fine and diffuse, and there is no distinction between the heartwood and the sapwood, making the timber smooth and even-grained. The wood is soft and lightweight. Once seasoned, basswood is very stable and free from both warping and shrinkage (Edlin 1969, 127). Basswood is not a very strong wood, but it has excellent working properties (Bramwell 1982, 259).

Beech was not mentioned in the literature as a wood for stretcher members, but the hard, strong, and heavy wood is used to make keys for stretchers. The even growth and small pores of beech make it a workable wood, and it has a smooth surface that resists wear (Edlin 1969, 99).

Cherry is a fine-textured and generally straight-grained wood. European cherry is slightly heavier in weight than American cherry. Once seasoned, cherry is moderately stable in use. It can be worked easily and gives an excellent finish (Bramwell 1982, 252). This wood is often used for keys.

Maple does not readily warp or shrink and is a very hard wood. The fine grain makes it easily worked. Maple timber (sugar or hard maple) is more commonly used in America (Gettens and Stout 1966, 240). This wood is not often used for stretchers, but it may be used for keys.

Poplar is soft, fine-grained, and fine-textured. The timber is light in weight and has a low tendency to warp due to its even grain (Gettens and Stout 1966, 254). It will decay in adverse conditions, but is a “tough [wood] for its weight and does not easily split or splinter” (Bramwell 1982, 252).

References

Ameringer, C. Seifen. 1999. Stretcher mounting/Re-mounting survey. Questionnaire results. Unpublished typescript.

Berger, Gustav A. 1980. Conservation of large canvas paintings: The role of constant tension mounting systems. Technology and Conservation 5 (Spring): 26–31.

Berger, Gustav A. and William H. Russell. 1988. An evaluation of the preparation of canvas paintings using stress measurements. Studies in Conservation 33:187–204.

Berger, Gustav A. and William H. Russell. 1994. Interaction between canvas and paint film in response to environmental changes. Studies in Conservation 39:73–86.

Berger, Gustav A. and William H. Russell. 1986. Investigations into the reactions of plastic materials to environmental changes. Part I: The mechanics of the decay of paint films. Studies in Conservation 31:49–64.

Berger, Gustav A. and William H. Russell. 1984. The new stress tests on canvas paintings and some of their implications on the preservation of paintings. ICOM Committee for Conservation, 7th Triennial Meeting, Copenhagen 84/2/7.

Booth, Peter. 1989. Stretcher design: Problems and solutions. The Conservator 13:31–40.

Bramwell, M., ed. 1982. The international book of wood. London: Artists House.

Buck, Richard D. 1972. Some applications of rheology to the treatment of panel paintings. Studies in Conservation 17:1–11.

Buck, Richard D. 1972. Stretcher design, a brief preliminary survey. In American artists' materials, Vol. II. A guide to stretchers, panels, millboards, and stencil marks. 1992. By A.W. Katlan. Madison, CT: Sound View Press.

Carlyle, Leslie. 2001. The artist's assistant: Oil painting instruction manuals and handbooks in Britain, 1800–1900, with reference to selected eighteenth-century sources. London: Archetype Publications.

Chase, William T. and Jeremy R. Hutt. 1972. Aaron Draper Shattuck's patent stretcher key. Studies in Conservation 17(1):12–29.

Cockerline, Neil. 1991. Plexiglass stretcher keys. Painting specialty group postprints. Albuquerque, New Mexico, June 8:56.

Conversations with painting conservators at the National Gallery of Art 1999–2004.

Edlin, H. L. 1969. What wood is that? A manual of wood identification. New York: Viking Press.

Erhardt, David and J. S. Tsang .1990. The extractable components of oil paint films, 93-97. IIC Contributions to the Brussels Congress Preprints.

Erhardt, David, Marion F. Mecklenburg, Charles S. Tumosa, and Mark McCormick-Goodhart. 1995. The determination of allowable RH fluctuations. WAAC Newsletter, Vol. 17, Number 1 (January):19–23.

Hagan, Eric, Alison Muray, Tom Learner, Christina Young, and Maria Charalambides. 2007. Factors affecting the mechanical properties of modern paints. Modern Paints Uncovered. Postprints of the Tate/Getty/NGA Symposium, May 2006.

Hedley, Gerry. 1993. Alan Phenix ed. The practicalities of the interaction of moisture with oil paintings on canvas. Measured Opinions. Ed. Caroline Villers. London: United Kingdom Institute for Conservation.

Hedley, Gerry, Marianne Odlyha, Aviva Burnstock, Jane Tillinghast, and Camilla Husband. 1990. A study of the mechanical and surface properties of oil paint films treated with organic solvents and water. Measured Opinions. Ed. Caroline Villers. London: United Kingdom Institute for Conservation. 1993, 103–111.

Hedley, Gerry, 1975. Some empirical determinations of the strain distribution of stretched canvases, 21–26. Proceedings of the ICOM Committee for Conservation, Venice.

Hedley, Gerry. 1981. The stiffness of lining fabrics: Theoretical and practical considerations. Measured Opinions. Ed. Caroline Villers. London: United Kingdom Institute for Conservation. 1993, 76–80.

Gettens, R. J., and G. L. Stout. 1966. Painting materials: A short encyclopedia. New York: Dover Publications, Inc.

Grimm, C. 1998. Personal communication with author.

Hoadley, R. B. 1980. Understanding wood: A craftsman's guide to wood technology. Newtown, CT: The Taunton Press, Inc.

International Council of Museums (ICOM). 1960. Report on fabric paint supports. Museum 13(3):135–171.

Katlan, Alexander W. 1992. American artists' materials, Vol. II. A guide to stretchers, panels, millboards, and stencil marks. Madison, CT: Sound View Press.

Katlan, Alexander W. 1999–2000. Personal communication with Sarah Fisher. December-January.

Kuschan, Edgar. 1999. Telephone conversation with author. 24 March.

Liu, Simon. 1999. Telephone conversation with Barbara Buckley. 5 March.

Mecklenburg, Marion F. 1991. Applied mechanics of materials in conservation research. Materials Issues in Art and Archaeology II, Materials Research Society Symposium Proceedings, Eds. P. B. Vandiver, J. Druzik, and G. S. Wheeler. Vol. 185:105–124.

Mecklenburg, Marion F. and Charles S. Tumosa. 1991. Mechanical behavior of paintings subjected to changes in temperature and relative humidity. Art in transit: Studies in the transport of paintings, 173–216. Ed. Marion F. Mecklenburg. Washington: National Gallery of Art.

Mecklenburg, Marion F., Mark McCormick-Goodhart, and Charles S. Tumosa. 1994. Investigation into the deterioration of paintings and photographs using computerized modeling of stress development. Journal of the American Institute for Conservation 33:153–70.

Michalski, Stefan. 1991. Paintings: Their response to temperature, relative humidity, shock, and vibration. Art in transit: Studies in the transport of paintings. Ed. Marion F. Mecklenburg. Washington: National Gallery of Art.

Quandt, Eleanor S. 1971. Technical examination of 18th century paintings. American painting to 1776: A reappraisal. Seventeenth Annual Winterthur Conference. Charlottesville: University Press of Virginia.

Russell, William H. and Gustav A. Berger. 1982. The behavior of canvas as a structural support for painting: Preliminary report. Science and technology in the service of conservation. Preprints of the contributions to the Washington congress, 3–9 September 1982, 139–45. International Institute for Conservation.

Straub, Rolf E. 1965. Spannvorrichtungen für das Leinwandbild. Nachrictenblatt der Denkmalpflege in Baden-Würtem-berg. 8(3):75–77.

Tara-Fredrix. 1999. Telephone conversation between representative and author.

Telephone conversations with manufacturer's representatives from the following firms: Archival Art Services, Washington, D.C.; Conservator's Stretchers by Wingrill, Ltd., Brighton, Ontario; and Simon Liu, Inc., Brooklyn, NY, February, 2000.

Von Sonnenburg, H. 1979. Rubens' Bildaufbau und Technik, I, Bildtrager, Grundierung und Vorskizzierung. Maltechnik Restauro 85:77–100.

Wingrill. Information pamphlets “Conservator's Stretcher,” examinations of sample stretchers. Mailed to author in 1999.

Young, C. R. T. and R. D. Hibberd. 2000. The role of canvas attachments in the strain distribution and degradation of easel paintings. Tradition and Innovation: Advances in Conservation. IIC Congress 2000 Preprints.

Additional Resources Consulted

Berger, Gustav A. and William H. Russell. 1990. Deterioration of surfaces exposed to environmental changes. Journal of the American Institute for Conservation 29:45–76.

Berger, Gustav A. and William H. Russell. 1991. The mechanics of deteriorating surfaces. Material Issues in Art and Archaeology II, Materials Research Society Symposium Proceedings. Eds. P. B. Vandiver, J. Druzik, and G. S. Wheeler. Vol. 185:85–92.

Carr, D. J., C. R. T. Young, A. Phenix, and R. D. Hibberd. 2003. Development of a physical model of a typical nineteenth-century English canvas painting. Studies in Conservation 48:145–154.

Cornelius, F. DuPont. 1967. Movement of wood and canvas for paintings in response to high and low humidity. Studies in Conservation 12:76–9.

Erlebacher, Jonah D., Eric Brown, Marion F. Mecklenburg, and Charles S. Tumosa. 1992. The effects of temperature and relative humidity on the mechanical properties of modern painting materials. Material Issues in Art and Archaeology III, Materials Research Society Symposium Proceedings, 359–70.

Hackney, Stephen and Gerry Hedley. 1981. Measurements of the ageing of linen canvas. Measured Opinions. Ed. Caroline Villers. London: United Kingdom Institute for Conservation. 1993, 57–65.

Hedley, Gerry. 1988. Relative humidity and the stress strain response of canvas paintings: Uniaxial measurements of naturally aged samples. Measured Opinions. Ed. Caroline Villers. London: United Kingdom Institute for Conservation. 1993, 86–96.

Hedley, Gerry, Caroline Villers and V. R. Mehra. 1980. Artists' canvases: Their history and future. Measured Opinions. Ed. Caroline Villers. London: United Kingdom Institute for Conservation. 1993, 50–56.

Hedley, Gerry and Marianne Odlyha. 1989. The moisture softening of paint films and its implications for the treatment of fabric supported paintings. Measured Opinions. Ed. Caroline Villers. London: United Kingdom Institute for Conservation. 1993, 99–102.

Karpowicz, Allan A. 1990. A study on development of cracks on paintings. Journal of the American Institute for Conservation 29(2):169–180.

Karpowicz, Allan A. 1989. In-plane deformations of films of size on paintings in the glass transition region. Studies in Conservation 34:67–74.

Karpowicz, Allan A. 1982. Some aspects of the mechanical behavior of fabric supported paintings. Report to the Smithsonian Institution. 12–15.

Karpowicz, Allan A. 1982. Some aspects of the mechanical behavior of fabric supported paintings. Unpublished manuscript.

Karpowicz, Allan A. 1988. The effects of atmospheric moisture on the mechanical properties of collagen under equilibrium conditions. AIC Preprints of papers presented at the Sixteenth Annual Meeting. 231–244.

Schilling, Michael. 1989. The glass transition of materials used in conservation. Studies in Conservation 34:110–116.

Verougstraete-Marcq, H. and R. Van Schoute. 1986. Painting technique: Supports and frames. PACT: Art History and Laboratory Scientific Examination of Easel Painting 3:13–34.

Young, Christina, Eric Hagan, Maria Charalambides, and Tom Learner. 2007. The interfacial interaction of modern paint layers. Modern Paints Uncovered. Postprints of the Tate/Getty/NGA Symposium, May 2006, in press.

Continue to Materials and Equipment Used for the Construction of Stretchers and Strainers.