The Abrasiveness of Sheer Overlay Fabrics Used in Textile Conservation

Fragile fabrics in textile collections are subject to deterioration due to use, exhibition, and improper storage conditions. Textile conservators often sew sheer fabrics as overlays directly over weakened fabrics to protect them from abrasion and to help maintain the integrity of the objects. Conservators rely on subjective opinions about fabric properties in choosing materials for their overlay treatments because 'objective data are not available. Textile properties, such as abrasiveness, of sheer overlay fabrics play a role in the success of conservation treatments over time. A survey of textile conservators provided data about the u,se of overlay fabrics including criteria for selection and type of objects being treated. Cross tabulation of the data revealed trends in the use of sheer overlay fabrics. Eleven fabrics were purchased from retailers. Properties, such as yarn type and woven or knit structure, were described, and eleven different textile performance tests were run. Nylon net was significantly more abrasive than polyester georgette and polyester English net. Three nylon nets were the sheerest fabrics. Other properties of sheer overlay fabrics measured in this research included cover, gloss, weight, thickness, surface roughness, coefficient of friction, elongation, electrostatic cling, and stiffness. Photomicrographs of fabrics and a summary table of specific fabric properties provide textile conservators with valuable information to use when selecting overlay fabrics.

role for museums and collectors by reducing the rate of this deterioration.
Conservators do not try to restore a textile object to a pristine or new-looking condition, but rather seeks to protect the textile from potential harm. They use cleaning techniques, stitching techniques, consolidation techniques, and control of the display and storage environments, among other things, to prolong the life of an object.
Modem fabrics as well as historic ones can be damaged by exposure to light, environmental pollutants, and changes in temperature and humidity, which affect the long-term stability of textiles. Conservators work to minimize these types of damage.
One method frequently used on fragile textiles that are fraying, splitting, or shattering is the use of a new sheer fabric sewn directly over the deteriorating fabric(s). The sheerness allows the color and pattern of the original fabric to show through, but keeps the damaged fabric intact. These protective fabrics are called sheer overlays and are used on weakened fabrics to protect them from abrasion and to help maintain the integrity of the textile objects.
Results of a survey done by this author, shown in Appendix A, show that conservators choose overlay fabrics for a variety of reasons including sheerness, fiber content, color, strength, hand, and availability. Conservators rely on subjective opinions about fabric properties in choosing materials for their overlay treatments because objective data are not available. Objective data would allow conservators to choose fabrics for technical reasons. Textile properties such as the abrasiveness of the sheer overlay fabrics play a role in the success of conservation treatments over time.
The sheer fabrics themselves may be detrimental and may contribute to an object's deterioration through abrasion rather than protecting the object during exhibition and storage. An evaluation of the abrasiveness of overlay fabrics will help conservators make an informed choice. This research studied the abrasiveness of typical overlay fabrics when abraded against a fiber donor fabric using non-accelerated test methods.
Relationships between the physical and mechanical properties of fabric are complex because of the multiple structural levels within fabrics. Fibers, yarns, and fabrics each have their own "geometrical and mechanical variables which control or influence to varying degrees the fabric behavior" (Pan 1996, p. 312). The objective of this research was to compare the abrasiveness of frequently used sheer overlay fabrics and to identify predictors for abrasiveness to be used when testing equipment is not available. After identifying the most typical overlay fabrics used by conservators today, a variety of fabric properties including fiber content, yarn type, fabric structure, weight, cover, gloss, stiffness, elongation, coefficient of friction, surface roughness, abrasiveness, and electrostatic cling were analyzed. These properties of the overlay fabrics were correlated statistically with the abrasiveness results as measured by digital image analysis of loose fibers on the overlay textiles after fabric-to-fabric abrasion. Statistical analysis of the data from the textile performance analyses provided insight into the interactions of the tested variables wi:th abrasion and with each other. The results ofthis research will provide textile conservators with valuable information to use when selecting overlay fabrics.
The ability to protect without harming a fragile textile is vital. Simpson (1993) did several studies looking at the ability of backing fabrics to protect without harm.
She commented that backing fabrics for conservation "should not introduce any circumstances that may weaken the historic textile" (p. 86). This also should apply to fabrics used as overlays. This research extends the Simpson research to sheer ov~rlay fabrics.
Conservators do not agree on the most appropriate fabrics to use as overlays for projects. Mailand and Alig (1999) recommended nylon net as an appropriate fabric for use as a sheer overlay. Landi (1998) suggested different fabrics for different uses including nylon net, Stabiltex, and silk crepeline. Blum and colleagues chose to use nylon bobbinet rather than Stabiltex or crepeline when conserving the Ormerod bedcover for its sheerness, dyeability, and width (Blu.m, Reiter, and Whelan 2000).
Many historic and fragile textiles on display in museums and galleries are composed of multiple layers. Some have hung in the same vertical position for many \ years. Gravity affects these textile systems, and an overlay added to the textile system by conservators adds one more potentially damaging variable to an object. If the components move at different rates, due to gravity or environmental conditions, and if any of the components are abrasive, damage will occur to the adjacent components.
Annis and colleagues studied "fiber transfer, the release and relocation of individual fibers from their original positions within a textile" (Annis, Bresee, and Cooper 1992 p. 293). Because abrasion in an exhibition or storage setting could be a slow but ongoing process, the ability to evaluate abrasion by single fiber transfer is important. This research used the single fiber transfer methodology developed by Annis et al. to evaluate the abrasiveness of sheer overlay fabrics. It also analyzed other mechanical and physical properties to study how fiber, yam structure, and fabric structure affect single fiber transfer. The other properties also were evaluated for use as potential indicators of a fabric's abrasiveness and for use by conservators when \.
selecting sheer fabrics as overlays.

Sheer Fabric Overlays in Conservation and Restoration
Textile conservators traditionally have chosen sheer fabrics such as nylon net, tulle, bridal veil illusion, bobbinet, silk crepeline, and polyester Stabiltex (also knoki as Tetex) as overlay fabrics to protect and support fragile and damaged fabrics from further deterioration during exhibition and storage (Ordonez 2001). Many historic costumes, flags, and bed coverings are made of silk fabric. Silk degrades with heat, light, excessive weighting, and age and often is found shattered in nineteenth-and twentieth-century textile objects. "The conservation of highly degraded shattered silk is problematic and as yet there is no entirely satisfactory approach to treatment" (Halvorson 1991p.4). Overlay fabrics often are used to protect fragile or damaged textiles made of silk and other fibers. Nineteenth-century dark brown cotton fabrics in quilts and costumes are another example of fabrics where sheer overlays are used.
These brown fabrics are frequently seen in a degraded condition due to the dyeing process they underwent in processing.
In a discussion of overlay fabrics, Lodewijks and Leene (1972) commented on the importance of sheerness, noting, "fabrics to be applied should be chosen to correspond with the object. A first requirement is that they should be as transparent as possible" (Lodewijks and Leene 1972 p. 142). Thomsen (1988) reported that use of silk crepeline continues in conservation "because of its receptivity to dyeing" (p. 33). Silk crepeline has the drawbacks of edges that fray easily and a weave structure that easily distorts while being manipulated. Nancy Kirk (2005) of the historic quilting website, The Kirk Collection, recommended silk crepeline for conservation but had a reservation about using it on frayed or broken fabrics on quilts noting that "it does not prolong the life of the original fabric, but does help keep it in the same plane as the quilt." Another quilt \. restoration specialist commented in the TEXCONS textile conservation Internet discussion forum that she did not like silk crepeline because it "masks the true colors of the textile ... by putting a fine haze over the piece." She went on to say that it does nothing to stop or slow the deterioration of the textile (Syler 2002). Mailand and Alig (1999) recommended the use of nylon net "over and behind a vulnerable hole or edge" to prevent fiber loss or damage in deteriorated textiles (p. 36). Rice (1972) recommended using synthetic fiber gauze such as nylon or a fine silk netting or crepeline as a protective sandwich for a fragile textile during wet or dry cleaning. In the case of a three-dimensional item such as clothing, he suggests a fine net bag made of nylon or cotton. Lodewijks and Leene (1972) also recommended a sandwich of fine "polythene gauze" when washing a fragile silk flag. He cautioned readers to choose the gauze carefully, as gauze with a small mesh such as polyester or silk crepeline would capture the larger particles of dust and dirt and prevent their removal. A coarse mesh might cause an imprint on the flag after drying. They also cautioned against the use of thick tulle fabrics in conservation because the ''texture may become imprinted into the thinner material of the object to be restored ... the hardness of the yarn will cause further wear of the restored object when it is handled" (Lodewijks and Leene 1972 p. 142).
Historically conservators chose overlay and support fabrics to closely match the fibers and yarns in the object according to Landi. More recently, synthetic fiber fabrics are gaining favor in the conservation community because of their supposed "greater resistance to environmental factors." She recommends that the ideal would be to match the reaction to environmental changes in the overlay fabric to the reactions of the textiles in the object (Landi 1998 p. 7 .2).
Sheer overlays of net may be used in quilt conservation, not to provide for short-term use by the current owners, but to maintain the history of the quilt into the centuries ahead (Wasserman 2002). Pampe (2002) suggested that sheer illusion or crepeline should be used to protect damaged areas during quilt restoration. In a 1987 Cooperative Extension bulletin on quilt conservation, Ordonez (1987) wrote that net or tulle may be sewn over weak or damaged areas for temporary support during wet cleaning. She no longer recommends this practice based on her own research and research done by Simpson on the abrasiveness of backing fabrics. 7 In 2000, conservators sandwiched the Ormerod Bedcover between nylon net and a cotton support as a part of its conservation in Philadelphia. They stated: "nylon bobbinet was chosen over crepeline or Stabiltex because it was the least visible." They noted a fuzzy sheen on the surface with crepeline and Stabiltex, especially when viewed from the angle at which the piece was to be displayed. Other positive attributes of the nylon net was its availablity in widths wide enough for the bedcover and it could be dyed in-house (Blum, Reiter, and Whelan 2000 p. 26). 8 In discussing crazy quilt restoration, Cognac (1994) commented: "some restorers of crazy quilts recommend netting. However, netting does not stop silk from flaking. Indeed, the netting can turn into pouch-like receptacles that hold the silk debris" (p. 44) She noted that netting can camouflage and obscure any embroidery on the quilt. In another chapter of her book, however, Cognac wrote that "fine crepeline, 1. tulle, or netting can be applied to the damaged sections of a crazy quilt" (p. 74) She suggested using different shades of net to enhance the fabric underneath and offered suggestions for attaching the netting with the fewest stitches possible. Quilt restorers are recommending the use of the newest nets to their students and clients without the knowledge or research to confirm that the nets are appropriate for the textile (Quilt-Restorers 2001;Wasserman 2002).
A Kansas State University Cooperative Extension booklet found on the web recommended using sheer fabrics to cover frayed or broken fabrics to "prevent further damage but allow the color to show through." The authors suggested using fabrics such as tulle, chiffon, and silk or polyester crepeline (Burke and Ordonez 1989 p. 5).

Flags
Flag conservation is another area in which conservators frequently use sheer overlays. The Museum of the Confederacy conservation program, in Richmond, Virginia, uses Stabiltex for flag encapsulation. Conservators sew around the perimeter of the flag and its fragments and in areas where material has been lost on the flag, taking special care not to sew through the fabric of the flag (Rawls 2002).
Conservation of a pair of British regimental flags was accomplished by "sandwiching them between two layers of nylon net" (Lennard 1995 p. 179). Stitching through the layers of net and not through the flags provided support. Lodewijks and leene also recommended a sandwich made of two layers of silk or polyester crepeline, dyed to match the deteriorated flags (Lodewijks and Leene 1972).
polyester Stabiltex for flag conservation when it became available. She had previously used silk crepeline but found that "fragility and sensitivity to light were limiting factors." The newer fabric also had drawbacks. Even though the Stabiltex is durable, strong, sheer, and available in a number of colors, the one-meter width is too narrow for many flags, making piecing necessary. The soft plain weave with no apparent sizing resulted in heavy fraying at the cut edges. The fabric's resistance to folding without heat setting makes encasing an edge difficult and leaves a strong colored line. Pollak and Thomsen (1991) reversed previous treatments on a Civil War era painted flag and then encapsulated the flag in custom-dyed blue Stabiltex. Two widths of fabric were hot-melt seamed to make a piece large enough to cover the flag.
They stated that the seam was visible as only a thin line. The fringe for the flag was separately encapsulated in gold-colored Stabiltex.

Conservation of the seven banners of St. Andrew's Church in Grafham,
Cambridgeshire, UK, included stabilization "by applying an overlay of adhesive coated silk crepeline, dyed to a sympathetic colour." The conservators chose overlays instead of underlays where the area of loss was not accessible from the back, and it was ''too fragile to withstand insertion of patches behind the weak areas from the front." They used silk fabric underlays on the areas of the banner with complete loss of silk in addition to the crepeline overlay (Townsend 1999).
In an article detailing a method for cutting a shape out of an overlay for a thickly embroidered section of an historic military flag, Dancause (2002) discussed the complex set of criteria used to choose a textile for a sheer overlay. The fabric needeCi to be colored, sheer, and strong and ''to complement the disparate elements composing the artifact" (p. l) It also needed to provide support to the weak ground fabric but have openings cut without fraying to allow the embroidered crests to stand above the ground without an overlay. The conservators chose to use Stabiltex and hot-melt cutting was developed to stabilize the edge of a cut without fraying.
Underlays of colored cotton, mimicking the flag's design, were used during restoration a Civil War era flag to minimize the visual disturbance of missing areas of the flag. The conservators then used hot-melt cut overlays of colored Stabiltex to further correct the color differences between underlay and flag (Pollak and Thomsen 1991). Sheer fabrics also can be used as supports for flags such as the Ocean Pond flag of the 61h Florida Battalion in the Civil War. The flag ''was hand-sewn to a single layer of polyester Stabiltex fabric for support" (How is a Flag Stabilized 2002). Lodewijks and Leene (1972) recommended a "dummy" flag made of crepeline or other thin material be placed over or under deteriorated flags "to give the impression of the original" but not "distract the observer's attention from the original fragment" (p. 174).

Costume
The textile conservator the Peabody Museum of Archeology and Ethnology reported that she stabilized the ribbons on a Micmac chief's coat by encasing the ribbons at the front panels and along the hemline in a "fine polyester ( crepeline-like) fabric" (Holdcraft 1998 an overlay on an Egyptian tunic at the Fitzwilliam Museum in Cambridge, UK, because it "was not visually intrusive, but clearly distinguishable from the original object on and off display" (Bacchus 1998 p. 13 ). Ekstrand (1972) reported on the conservation of a black silk ribbon, which supported the standing brim of a hat, by stitching it to "a strip of selvedge of tulle" (p. 194).
Archeological Textiles Negnevitsky and Schick (2000) chose to use two different sheer fabrics to conserve a large linen burial cloth in Israel, based on specific features: "Stabiltex and silk crepeline were selected. The former, a pure synthetic, is the stronger of the two--a positive feature; but the latter is the more transparent and less glossy-a positive feature for display and study" (p. 147). As a part of the treatment, the 6000- year-old textile was sandwiched between Stabiltex and silk crepeline.

Household Furnishings
In the Care and Preservation of Textiles, Finch and Putnam (1985) stated that using a net or crepeline cover over furnishing fabrics "will not prevent actual deterioration, but it will ensure that any potentially loose threads are kept in place and not rubbed away and lost" (p. 265). She noted that this provided the "least possible chance of the fragile fibers being broken, twisted, or pulled out of shape" (p. 265).
\ She also recommended the use of net coverings over hooks, eyes, and snaps to protect the rest of the object from damage during wet or dry cleaning.
When wall hangings at Ham House, Surrey, UK, were conserved, the staff discovered that the back of most panels had been adhered to a silk tulle net support.
The patterns of the net underlay created impressions on the front of the damask in several places. The conservators noted that those panels adhered to net had greater warp face breakdown than those not adhered to net. The breakdown of the damask warp face had been exacerbated by the rigidity of the adhered net backing. During the 1990 treatment, conservators chose to use Stabiltex overlays because of its durability and resistance to photo-degradation and chemical attack. They commented: "Stabiltex is denser in appearance than nylon tulle, but its visual appearance is quite acceptable when used vertically" (Hillyer 1990 p. 187).
Conservators prepared an historically important curtain at Uppark, West Sussex, UK, that had been damaged in a house fire for long-term storage by sandwiching it between "two layers of dyed conservation net." Stitching around the burnt fragments and along the festoon lines held the net in place. They encased the fringe separately in net and rolled the entire package on an acid-free tissue-covered PVC roller for storage (Marko 1995 p. 114). A tester cloth with a large burn hole from the Uppark state bed, damaged in the same fire, was treated with a piece of dyed nylon net over the damaged area and stitching beyond the margins of the hole into strong areas of the damask (Singer and Wylie 1995).
Dyed net, cut to fit, was the overlay of choice for faded historic silk wall hangings at Arlington Court, Barnstable, Devon, UK. The object of this conservation was to "even out the visual differences between faded and unfaded silk and to protect the silk from the light and damage caused by visitors" (Hutton 1995 p. 168). The conservators concluded that the net improved the visual appearance and perhaps provided a barrier between visitor and silk wall covering. They were not sure that the net would provide any protection from photo deterioration, although they took other measures such as keeping the drapes closed to control light levels. In addition, they stated, "although many of the larger degraded areas were supported, the net did nothing to support the overall weakness of the hangings" (Hutton 1995 p. 170).
A conservator used dyed nylon net to conserve chair seats at Felbrigg Hall, Norwich, Norfolk, UK. She secured the net under the back seat rail of the chair and gently tensioned it to the front. She tucked the net inside and secured it with lines of support stitching. She also used an underlay in this treatment, and the conservator felt that "the weak and damaged area of silk was well supported by the combined patch support treatment and the nylon net covering" (McClean 1995 p. 184).
Daily control of dust in historic houses and museums is essential. Lloyd (1995) suggested that covering the upholstery tool of a typical vacuum cleaner with fine nylon net and vacuuming directly on the upholstery surface was an appropriate way to clean furnishings. Lodewijks and Leene (1972) recommended the use of tulle as a screen over the mouthpiece of a vacuum cleaner prior to vacuuming historic textiles and furnishings. Owen (1995)  After the paint was dry, they removed the Mylar and stitched the overlay into place along the long edges and through some of the original quilting stitches. The curator of Hampton House was satisfied with the final appearance of the quilt and agreed to allow only limited exhibition of the quilt in the future (Schmalz 1999).

15
Environmental changes, such as variations in temperature and humidity, occur in exhibition spaces where textile objects are on display. Such changes might be stressful for an object already weakened through age and deterioration. Storage in ideal conditions may not cause stress to an object, but not all objects are stored in ideal or consistent conditions. Handling during examination, photography, conservation, and exhibition all put strain on an object. Movement between environments for these tasks is potentially damaging. "Early restoration or conservation treatments may have caused deterioration, even though most of them were carried out with the best of intentions" (Pye 2001 p. 93).
Stretch Due to Gravitational Forces Simpson (1991), based on her research on the abrasiveness of backing fabrics in conservation, suggests that a rough support surface is detrimental to a fragile textile on display due to its cutting into the delicate fibers and fabrics. She noted that this is a particular problem with vertical display or storage of a textile that allows gravitational force to exert pull on a historic textile against a backing fabric. Changes in temperature and humidity also cause movement of the two fabrics against each other.
Simpson also states that "even general handling of the item such as lifting and \ repositioning may cause movement of the two fabric surfaces against each other" (p.

179).
In  Townsend (1999) reported that gravity and environmental conditions in the St.
Andrews Church, Grafham, Cambridgeshire, UK, over a period of 100 years caused the upper edge of all the decorative banners to fall into scallops and hang in undulations between the loops that supported them. In a report about the conservation of a Civil War era flag, Pollak and Thomsen ( 1991) noted that the vertical stitches connecting layers of fabrics were not secured at the bottom of each stitching row "to allow for possible movement of fabrics in the flag as it was displayed" (p. 15).
Dimensional Change Due to Temperature and Humidity Many textiles, especially those made from cellulosic and protein fibers, expand and contract with changes in humidity. The resulting changes in size and morphology result in movement and friction. "Physical damage can be caused by juxtaposition of materials which expand and contract at different rates and to different extents" when exposed to changes in temperature and humidity as exemplified by paintings on linen \.
canvas on wood stretchers (Pye 2001 p. 90). Changes in relative humidity (RH) affect organic objects by changing the rates of chemical reactions and affecting the physical properties such as size, strength, and stiffuess (Erhardt and Mecklenburg 1994). A condition report made prior to conservation of the seven banners belonging to St.
Andrew's Church stated that "the appliqued fabrics appeared wrinkled in places, suggesting that the foundation linen had shrunk due to the environment of the church" (Townsend 1999). Berger and Russell's (1990) research on the responses of canvas paintings to controlled environment changes is relevant here. They found that "every canvas measured ... responded to changes in relative humidity and temperature ... and these responses often changed gradually during testing" (p. 2). The responses of stretched canvases varied greatly. For example, the tension in one canvas (linen, light weave, commercially sized, acrylic primed) rose with a rise in relative humidity while a second canvas (linen, basket weave, commercially glued and primed with oil paint) showed peak tensions during periods of low relative humidity. Individual canvases had a wide range of tension values in response to combinations of relative humidity and temperature, and generalizing about the variations was difficult. This may be due to different sizing treatments, types and styles of stretchers, type of paint, and painting technique. It also might be due to differences in the linen yams used to weave the canvas and even the sett, weave structure, and finishing of the canvas fabric itself. Berger and Russell (1990) found that exposure to high relative humidity produced a rise in tension in the canvas and permanently deformed and enlarged the canvas. He noted that when a fabric loses its tension, it also loses its elasticity and its resistance to deformation. "Once stretched out, canvas can never be pushed together 'again by mechanical means" (p. 4). Cycling environmental changes caused by intermittent exposure to lamps or reflectors in a gallery situation also are important considerations. Erhardt and Mecklenburg (1994) state, "moderate changes in relative humidity produce minimal problems in materials that are free to expand and contract." They go on to say that larger changes cause problems even for those objects free to expand and contract because the rate of moisture diffusion takes time and different materials have variable rates, causing swelling or contraction of one part of an object with potential damage to other parts. Padfield's (2003) research indicated that ''temperature change, typically between ten and twenty degrees, caused more change of stress than a RH change between 15% and 55%" (p.2). This is a serious challenge to the prevailing orthodoxy that conservators can be relatively careless about the temperature but must insist on a steady relative humidity. Furthermore, the reaction of paintings on fabric to small, rapid temperature changes was disproportionately large compared to changes in humidity alone. The effect of the rate of change of environmental variables is a subject that looms large in the field of conservation. Berger (1981) observed the good condition of a set of gigantic Atlanta Cyclorama dioramas that traveled around the USA in the last century. These pictures were rolled up for transport, between being exhibited hanging from a hoop. Another hoop, loaded with weights, held the bottom under tension. The bottom, therefore, was free to rise and fall with the changing climate. The variable waistline of the diorama allowed the shrinkage in the middle of the canvas to relieve the horizontal stresses.
The affect of moisture on abrasiveness and abrasion resistance is complex.
Moisture serves as a lubricant, reducing the friction between a fabric and another surface, potentially slowing the abrasion process. Damp, swollen fibers are also less brittle. Many factors must be considered when observing the interactions between environmental conditions and abrasiveness. For example, fabrics made of fibers that are stronger when wet than dry have better resistance to wet abrasion than to dry abrasion, and conversely, fabrics of fibers that have lower tensile strength when wet than dry may abrade more easily when wet (Collier and Epps 1999). Barring disaster, textiles on display and in storage may not be thoroughly wetted, but changes in relative humidity will affect the abrasiveness and abrasion resistance of any two fabrics in contact with each other.

Airborne Pollutant Damage
Airborne pollutants, which can settle on any unprotected surfaces of textiles and objects, can cause discoloration, mold-growth, and abrasion. Airborne pollutants include sulfur and nitrogen oxides that can react with water to form acids and ozonean oxidant. Particulate pollutants include dirt, and building and display materials such '· as concrete dust, wood, wood composites, adhesives, textile fibers, and textile finishes that can all release alkaline particles, organic acids, formaldehyde, or hydrogen sulfide. Dirt particles can penetrate into porous objects, while oily pollution will sit on the surface. Airborne dirt includes gritty particles, tiny bits of skin, carbon particles, oil droplets, bacteria, and mold spores (Pye 2001). Pollak and Thomsen (1991) report that a Civil-War-era painted silk flag brought to them for conservation had suffered from a 1970 conservation treatment. It was pressure mounted in a painted pine frame and covered with two sheets of Plexiglas with 1/2 inch gap between the two adjacent sheets. The gap allowed airborne pollutants to reach the flag. In 1970 the flag fragments were glued to a silk crepeline backing with a water-soluble adhesive. An overlay of silk crepeline was added, and the entire sandwich was machine-stitched with synthetic mono-filament thread over the entire surface of the flag. The adhesive was washed out, and the flag ironed "to set the stitches and further flatten the flag." By 1990, the conservators noted that the "previous treatment was causing continued stress and damage to the flag." The silk crepeline overlay had been dyed blue to match the flag, but in 1990 it was a "dull grimy blue-grey color that greatly obscured the design" (pps. 10,12). The crepeline had become brittle and no longer provided support or protection for the flag.

Pressure-Mount Framing
The Museum of the Confederacy stores its flag collection flat in a new storage facility. Flags destined for exhibition are laid on an unbleached cotton panel over acid-free board padded with polyester batting. If a flag has been encapsulated in Stabiltex, it is sewn to the backing fabric through the sheer fabric. For exhibitions the conservators cover the flags with ultraviolet-blocking Plexiglas and add a custommade aluminum pressure mount frame to apply light pressure to the entire sandwich (Rawls 2002). Pressure mounts are designed to prevent movement of the textilemount sandwich layers. The pressure might be strong enough to cause the strong synthetic yams of the overlay to cut into the silk flag. "Sometimes a small degree of friction is helpful in holding the two textile items together so that they do not slide 'apart easily" during display (Simpson 1991 p. 179). The long-term effects of pressure mount framing have not been studied.

Studies of Abrasiveness of Textiles Used in Conservation
Crockmeter Abrasion Tests Simpson (1991)  she counted the numbers of fibers loosened from the flannel during abrasion by the four unbleached, 100% cotton support fabrics. She found statistically significant differences in abrasiveness between the four fabrics. Fabric construction, including weave structure and weight, had a significant effect on the amount of fiber removed from the flannel fabric. Of the tested fabrics, only the sateen weave showed a difference between the face and back of the fabric. Simpson's methodology did not replicate actual conditions of storage or display, but it did provide a "procedure for measuring fiber loss due to abrasive action of fabric surfaces" (p.91). Simpson (1991)'selected the simple method of counting the number of loosened fibers and fiber particles transferred from one fabric to the other during abrasion using a linen tester for magnification to evaluate the fabric-to-fabric abrasion. Simpson ( 1993) continued her investigations of fabric-to-fabric abrasion and fiber transfer with eleven new fabrics. She again used a 100% cotton napped flannel as the "fiber donor fabric" and the crockmeter to provide the low-force abrasive action. Statistical analysis showed that the eleven fabrics could be grouped by their abrasiveness with tulle being the most abrasive and the group consisting of muslin, velveteen, and two weights of laminate as the least abrasive. Inverse relationships for weight and thickness compared to abrasiveness resulted for tulle and velveteen, but for simple woven fabrics with similar abrasiveness values for both face and back, the heavier and thicker the fabric became, the more abrasive it was. Simpson (1993)  yarn, and fabric stages can be analyzed using low levels of abrasive force (Annis 1990). The analysis of fiber transfer can be used to detect abrasion damage as limited as the loss of a single fiber fragment. They define fiber fragment as a short piece of fiber broken horizontally across the fiber, not as longitudinal fibrillation. This research also uses this definition.
Forensic science research has identified three basic mechanisms of fiber transfer: shedding of loose fibers residing on a textile's surface, disentanglement and removal of whole (unbroken) fibers partially embedded in a textile' s interior, and fracture of whole fibers followed by the release of fiber fragments (Pounds and Smalldon 1975). Fiber transfer plays a part in most textile wear mechanisms. All of the research by both Simpson and Annis used single fiber transfer analysis to observe and measure all three mechanisms when they happen in a controlled setting. The 'machine developed by Annis and Bresee (1990) (Annis, Hsi, Bresee, and Davis 1998). Fuzz forms when fibers are fractured and pulled from yarns producing lint or fuzz on the fabric surface. They measured fuzz as the height of the entangled fibers above the fabric surface. Further research in this area resulted in an understanding that the yarn interlacement pattern influences the amount of frictional energy transferred from one fabric to the other and accounted for the differences in fuzz formation . Backer and Tanenhouse (1951) observed the differences in abrasion performance "when the direction of rubbing is altered with respect to warp and filling coordinates" (p 648). Annis and colleauges also showed that more and longer fuzz was produced by orbital abrasion than by linear abrasion (Annis, Davis, Bresee, and Hsi 2001). During fabric-to-fabric abrasion, distinguishing the fibers from each fabric is essential. Annis  Simpson, to count the number of transferred fibers. Counting systems, combined with removing and measuring fibers longer than 2 mm, also were tried and achieved results that included the number of fibers, mean length of fibers, and a length distribution (Annis, Bresee, and Cooper 1992). Ultimately, Annis chose to use microscopic video imaging to compare fabric surface changes qualitatively (Annis, Bresee, and Warnock 1991 ). Digital photo-micrographs allowed for various digital measurements and comparisons of the individual fibers (Annis 1990). Simpson (1991) suggested that weighing on a very accurate scale could evaluate the amount of fiber transferred. In later research, Hsi, Annis, and Bresee (1998) reported on improved image analysis for analyzing the fuzz left on fabrics after non-accelerated abrasion. This method of analysis required specialized hardware, software, and trained personnel.

Research on Physical and Mechanical Properties of Fabric
Physical Properties of Fabrics Many properties of fabrics contribute to their suitability as overlays. These include cover, luster, gloss, construction, finish, abrasiveness, friction, stiffness, roughness, elongation, and electrostatic cling. The following discussion explores these properties.

26
The amount of cover provided by a sheer overlay is critical to the success of a treatment. The definition of cover is ''the ratio of fabric surface occupied by yarn to the total fabric surface" (Kaswell 1963 p. 450). Cover is important when considering light, moisture, or water vapor penetration. Cover can be considered a measure of transparency because transparency implies the passage of light through a textile.
Transparency and sheerness often are linked in the discussion of fabrics. The term sheer is defined in Fairchild's Dictionary of Textiles as "transparent or lightweight fabric such as sheer chiffon, crepe, georgette or voile of various constructions" (Tortora and Merkel 1996). Based on fiber content, a textile can be transparent without being sheer, such as a fiberglass curtain. Conservators use sheer overlays for their transparency and associate transparency with lightness of weight and hand. In textile conservation, light permeability of a sheer overlay affects the ability to see the color and/or pattern of the underlying fabric through the overlay. Discussion of gloss, weight, and hand are considered elsewhere in this paper.
Grosberg states that the "shape taken up by the yarn in the warp or weft cross section of the cloth" may be important in assessing the relative resistance of cloths to the passage of air or light" (Grosberg 1969 p. 323). Fabric made of the same fiber, but differing in fabric count, ratio of warp and weft fabric count, ratio of yam spacing,' " average of yam spacing, and/or weave structure may still be "compared for one simple property that is the proportion of the total area of the cloth that is covered by the yam" (Grosberg 1969 p. 334). Two different fabrics may be considered similar if they "possess the same fractional covering power" (Grosberg 1969 p. 334). For this research, cover is considered a measure of transparency and is called sheerness because of the convention of conservators calling the fabrics they use sheer overlays.

Reflectance, Including Luster and Gloss
Visual characteristics, such as luster or gloss, affect how a sheer fabric looks when placed over a fragile textile. Luster is defined as the "amount of light reflected from the surface of a fiber, yam or fabric," while gloss is defined as the "luster or brightness of a fabric ... in a specific direction" (Tortora and Merkel 1996 pp. 336-7).
A lustrous or glossy fabric overlay might detract from the matte finish of an historic object. The addition of titanium dioxide (Ti0 2 )to the polymer melt of manufactured fibers in various amounts affects the intensity of fiber luster, sometimes called the brightness of the fiber. The type of fiber or filament and the amount of spin of a yam affects luster. Filaments are smooth, and this gives them "more luster than spun yams, but the luster varies with the amount of twist in the yarn. Maximum luster is obtained by the use of bright filaments with little or no twist" (Kadolph and Langford 1998 p. 220). Luster also may affect the abrasiveness of the overlay fabric, because "variations in the Ti0 2 content also affect the geometry of the fiber surface, namely, bright polyester fibers have a smooth surface, whereas dull fibers have a rough surface" (Schick 1977 p. 49).
Industry uses a measurement of gloss to "compare and match different components and to develop new effects for textiles" (Breugnot 2004). The overall ability of an object to scatter light defines its visual appearance, while the specific color and the gloss of an object combine to create its visual aspect. Measurement of those two important parameters is crucial to control the quality of the visual characteristics of an object. Often, especially for textiles and garments, the object to be analyzed is non-planar and is a complex three-dimensional surface (Kato Tech 2004).
Gloss measurement takes both scattered light and reflected light into account.
Because of this, the measurement correlates to human visual observation. A gloss meter evaluates gloss, color-codes it, and then using image processing, creates a mean value for gloss. Gloss measurements are appropriate for textiles because "light scattered by a rough surface is composed of diffused and specular components. The gloss degree of a surface is the proportion of specular reflection compared to the diffuse reflection at the surface whereas the color information is fully contained in the purely diffused light" (Breugnot 2004). Currently, no gloss standards exist for specific fibers or textiles.

29
Luster can also be defined as the directional variation of reflectance. This means that a glossy or lustrous fabric has the capacity to reflect more light in some directions than in others. Several companies have developed gloss meters that measure luster. In research investigating the effects of weave structure, fiber content, and yarn twist on luster, Kim and Shin (2004) concluded that as yarn twist in multlfilament yarns increases the luster unit size diminishes, which results in a macro-level gloss decrease.

Knitted Fabric Structures
Conservators use warp knitted fabrics as sheer overlays (see Appendix A).
Warp knitted structures have "unique properties of form-fitting and elastic recovery based on the ability of knitted loops to change shape when subjected to tension" (Spencer 1989 p. 248). Dimensional changes also can occur during use, and problems of shrinkage, stretch, and shape or size distortion affect consumer satisfaction with the fabric. One of the basic laws governing the behavior of knitted structures (as defined by the Hosiery and Allied Trades Research Association in the UK) is that "loop shape determines the dimensions of the fabric and this shape depends upon the yarn used and the treatment which the fabric has received." Loop lengths within each course affect fabric properties such as weight (Spencer 1989 p. 249).
Fabrics knit from synthetic thermoplastic yarns such as nylon and polyester can be heat set to a shape or dimensions. Hydrophilic fibers, such as cotton and silk, often show dimensional change after knitting, and this can cause major problems for the end user. Spencer (1989) wrote "in theory, knitted loops move towards a threedimensional configuration of minimum energy as the strains caused during production are allowed to be dissipated so that ... a knitted fabric will reach a stable state of equilibrium with its surroundings" (p. 255). Environmental conditions such as temperature and relative humidity will affect this state of equilibrium as will the mechanical properties of the fiber, yarn, and knit structure. \.
The structure of knitted non-knotted nets is achieved through manipulation of the standard knitting process. "Symmetrical nets are produced when two identicallythreaded guide bars overlap in balanced lapping movements in opposition" (Spencer 1989 p. 297). A hexagonal mesh, marketed as tulle by the industry, is produced by an open lap followed by a closed lap that causes the lapping to alternate between two adjacent wales and forms underlaps and inclined overlaps that close the top and bottom of the staggered mesh holes (Spencer 1989).
Tricot fabrics are the lighter end of warp knits-usually less than 4 ounces per square yard or 140 grams per square meter for the apparel and furnishing categories (Thomas 1976). The machines that produce tricot fabric use spring-beard needles.
Rasche! machines use latch needles. Many raschel machines make laces, plain nets, and elastic nets (Thomas 1976). In addition, according to Spencer (1989) the raschel machine is "more suitable for utilizing synthetic filament yam than traditional lace machinery" (p. 311 ). In 1976, when Thomas was writing, he noted, ''the boundaries between raschel and tricot fabrics are becoming ever more indistinct" (p. 41 ). The intervening years have almost certainly accelerated this process. Even so, differences in the physical and mechanical properties of raschel and tricot knit sheer fabrics may affect the abrasiveness of those fabrics and their effectiveness as overlays in conservation.

Fabric Finishes
Fabric finishes also must be considered when conservators choose fabrics for \.
sheer overlays. The behavior of knitted structures depends upon the yarn used and the treatment the fabric has received during production, according to Thomas (1976).
This is also true of woven fabrics. These treatments may include fabric finishes to change the appearance or performance of the textile and apply to wovens as well as knits. Some finishes are temporary, applied to facilitate production and construction of the final product. Others are applied to enhance the hand or appearance until the consumer makes the purchase and then may be removed with the first cleaning.
Manufacturers call these counter finishes (Schindler 2004 ). Chemical modification or thermosetting makes other finishes permanent. Finishes that are expected to last the lifetime of a garment are called durable. Collier and Tortora (2001) noted that information about these appearance-enhancing finishes is not usually included on a product label. Finishes applied to affect fabric performance are more likely to be chemically applied than those that affect appearance (Schindler 2004).
Finishes may soften the hand of a fabric or stiffen it. Manufacturers use finishes called hand builders to provide stiffness and added fullness to a fabric. These work because they attach to the fabric surface and accumulate in the spaces between Yarns. Individual fibers and yarns are bound together by the finish to create stiffness.
Such finishes are likely to affect properties other than stiffness such as stretch, friction, and surface roughness of a fabric (Schindler 2004). Many of the finishes affect more than one property of a fabric. For example, methacrylates are primarily hand builders, but they also can improve abrasion resistance, adhesion, elasticity, flexibility, and water or solvent resistance depending on copolymerization with other acrylic and vinyl monomers (Schindler 2004).
Conservators need to know which finishes have been used on the fabrics they \ use for overlays especially because chemical finishes can deteriorate over time and cause changes in the textile. For example, formaldehyde-containing thermosetting polymers tend to "reduce abrasion resistance, yellow after exposure to heat and release formaldehyde" (Schindler 2004 p. 84). Schindler noted other possible draw-backs to hand-building finishes such as "increased soiling and staining of finished fabrics, and increased fabric flammability" (Schindler 2004 p. 92). Unfortunately, retail labels and fabric retailers seldom have information about finishes, and manufacturers do not reveal their formulations of fabric finishes.

Mechanical Properties of Fabrics
Abrasion Resistance and Abrasiveness The purpose of conserving fragile and historic textile objects is to slow deterioration of the object. Placing two textiles in close proximity, without complete knowledge of the overlay textile's characteristics, may actually contribute to the deterioration through abrasion. Collier and Epps (1999) defined abrasion as "the mechanical deterioration of fabric components by rubbing against another surface" (p.
128). Booth ( 1969) defined abrasion as "a series of repeated applications of stress" (p.
296) adding that fibers held firmly by tension, pressure, or high fabric count will suffer more stress than those held only lightly. Warfield et al. (1977) agreed with Steigler et al. (1956) that "frictional abrasion has been found to be one of the causes of appearance degradation during use as well as a contributory cause of fabric failure in specific end uses" (Warfield, Elias, and Galbraith 1977 p. 332). Simpson (1993) stated that conservators should avoid placing "rough surfaces adjacent to the historic textile item" because the "rough surfaces of backing fabrics may abrade delicate fibers in the historic textile" (p. 86). "Abrasion ... affects the appearance of the fabric" (Collier and Epps 1999 p. 128). Simpson's research made conservators aware of the abrasiveness of backing fabrics, but conservators also need to understand the abrasiveness of sheer overlay fabrics (Simpson 1991(Simpson , 1993. Fabric-to-fabric abrasion occurs whenever fabrics are in use. The wearing of clothing provides the most obvious example, but any multi-layer textile in a museum or storage situation can experience fabric-to-fabric movement due to physical handling, gravitational pull, and temperature and humidity changes. "Although these types of fabric-to-fabric rubbing usually do not involve a significant amount of force individually, eventually fabric abrasion will become noticeable when they occur repeatedly, particularly when other types of abrasion occur simultaneously" (Collier and Epps 1999 p. 130). Abrasion also can occur when a textile is in contact with another surface such as a wall, abrade. This is especially important where bent or flexed edges such as folds or creases are rubbed against other surfaces.

34
Another consideration when evaluating potential abrasiveness is that "particles of dust, sand and other foreign substances held within the fabric can abrade yarns and fibers" (Collier and Epps 1999 p. 131 ). This is called third-party abrasion. In historic textiles, such as flags, third-party abradants such as environmental pollutants and salt from sea air and water can be very detrimental to the stability of the fabric structure.
Past treatment of a textile with soluble salts, such as the iron or tin salts used in the weighting of silk, can introduce a third-party abradant and also cause yarn and fiber deterioration. "In actual use many different abradant forces also act on a fabric at one time, while most laboratory tests simulate only one type of abrasion" (Collier and Epps 1999 p. 138).
Most research on textile abrasion has investigated abrasion resistance, not abrasiveness. Saville (1999) commented that test results regarding the factors that affected abrasion resistance in fabrics are contradictory largely because the tests were carried out using ''widely different conditions and in particular using different modes of abrasion" (pp. 195-6). However, these abrasion resistance tests provide insight into the reactions of fibers, yarns, and fabric to abrasion.
After one of the standard abrasion tests, accelerotor abrasion, Warfield and Stone observed that changes included fiber debris, voids, displacement of yarns, and pills for both polyester and cotton fabrics (Warfield and Stone 1979). Specific characteristics of the fiber and yarn play a more important role than just fiber type.
Filament yarns are more abrasion resistant than staple yarns because removing elements from filament yams is difficult. High-twist yams are more abrasion resistant than low-twist yams, again because of the relative ease of removing fiber elements from the low-twist yarns (Saville 1999). Low-twist yarns also have the ability to distort or flatten under abrasive pressure, and this allows them to be more abrasion resistant. This ability to distort may cause them to be less abrasive because they can conform to the shape of the other fabric. Nylon is highly abrasion resistant due to its high elongation and elastic recovery; polyester also has good abrasion resistance for \ these same reasons (Saville 1999).
Fabric structure also plays an important role in abrasion resistance (Saville 1999). Warfield and Stone (1979) pointed out the importance of fabric geometry "in the translation of inherent fiber properties" to the final fabric (p. 251 ). Woven structures with floats of relative mobility absorb the stress of abrasion better than those without such mobility (Saville 1999). Up to a point, a fabric with a higher fabric count is more abrasion resistant than a lower one, but then, as yarns become too packed to move within the fabric structure, resistance goes down. Warfield et al. (1977) found that tighter, more compact weave structures held fiber ends in the fabric better and inhibited fiber release, breakage, and pill formation.
Becker and Tanenhouse suggested that a firm binding of the fibers through either increased yarn twist or tighter weaves minimizes fiber release. They demonstrated this in wear testing: as twist of either warp or filling yarns increased, the abrasion resistance increased (Backer and Tanenhaus 1951). Warfield et al. (1977) also showed that fabrics with increasing amounts of yarn crimp had more abrasion damage than those with lower crimp. Increases in yarn crimp were associated with increases in filling thread count. Despite numerous tests on a variety of fabrics, the researchers were unable to designate any one fabric as the "best performer in both appearance and performance categories" (pp. 333-40). They noted that neither microscopic observation nor any single physical test was able to adequately evaluate the effect of frictional abrasion on the fabrics tested. Backer and Tanenhouse (1951) also studied the relationship between the structural geometry of textiles and abrasion resistance. They concluded that better abrasion resistance could be achieved by increasing the geometric area of contact between the fabric and its abradant. An increased number of warp crowns per square inch reduced the normal load per warp crown and increased the abrasion resistance.
Normal load is the weight of an object before any force is applied. Increased warp textures (at constant picks per inch) and increased filling texture "resulted in increased fabric cohesion and greater fabric durability" (pp. 636-40). Reducing the flexibility of the fabric by jamming in additional yarns decreased the fabric's durability. They observed that spun yarns in a fabric without a finish had radically altered yarn structure, post abrasion, due to surface fuzz of ruptured fibers. Increasing fabric thickness and larger yarn diameter increased abrasion resistance, although the relationship between thickness and yarn diameter is complex.
Temperature and humidity changes affect the size and shape of textiles, but they also may change the yarn friction and therefore the abrasiveness. Schick noted that the friction of cotton and rayon yarns increased with increasing moisture regain.
He suggested that the increase in friction is due to "an increase in area of contact. .. by the swollen rayon yarn" (Schick 1977 pp. 26, 30, 31 ).

38
Fabric-to-fabric friction is one of the factors creating abrasion between a sheer overlay fabric and an historic fabric. Ordinarily, researchers measure friction as "the force that resists the movement of an object." Static friction is the force needed to initiate movement; dynamic friction is the force required to keep the object in motion.
The phenomenon of friction is governed by a set of laws that "hold fairly well for hard materials, but not for textile materials particularly at low values of normal force" (Saville 1999 pp. 110-11 ).
During friction testing, Qiu et al. reported that the resulting coefficient of friction values are strongly dependent on test conditions (Qiu, Wang, and Mi 1999).
"Since a change in the angle of contact causes a proportional change in area of contact. .. friction is proportional to the area of contact" (Schick 1977 p. 24).
Therefore, if the area of contact in the testing conditions is different from the area of contact in actual conditions, results of measuring of the coefficient of friction may not be applicable to actual conditions.
Many studies have focused on frictional properties of yam for spinning and weaving operations. Fabric friction is subject to the same rules as yam friction according to Saville (1999). The angle of contact with the surface and the tension at either side of the contact affects yam friction. Increasing the angle of contact increases the frictional force due to "an increase in the normal force rather than to the increased area of contact." The frictional force can be kept constant for increasing areas of contact by keeping the angle of contact constant through increasing the radius of the contact surface (Saville 1999 pp. 111-12). These rules of contact may be applicable for the conservation of textiles, especially for those being rolled for storage.
The three "basic factors which determine the frictioanal resistance to mechanical deformation" in textiles are: 1) the ratio of the relative yam or fiber movement to the fabric deformation as determined by yam and fiber geometry; 2) the force between yams and fibers at intersections, which is largely determined by the state of unreleased manufacturing stress in the fabric; 3) the coefficient of friction between yams and fibers, which is dependent on the fiber type and surface characteristics as well as the presence of 'softening' agents or other lubricants or finishes (Schick 1977 p. 572). Olofsson and Gralen (1950) found that small changes in the fiber surface such as the height and shape of scales on wool yams produced a large change in the coefficient of friction in fiber-to-fiber friction. Measurements of the coefficient of friction are specific for the two materials in contact with each other, so measuring the friction of each fabric against a standard surface does not necessarily correspond to the friction a fabric will exhibit against another textile surface. This is unportant to consider when comparing results from two or more studies.

\.
Ajayi (1992) studied the effects of fabric structure on frictional properties in fabric-to-fabric friction. His research showed that the "knuckles" or crowns formed by the crossover of yarns in the two rubbed fabrics became engaged and thus restricted relative motion. He summarized by saying that ''the frictional properties of woven fabrics may be interpreted in relation to surface smoothness and texture from the geometric consideration of their component yarns." Ajayi showed that the "frictional resistance to motion increased as the relative area of contact (fabric balance) between\ the fabrics increased." He also found a similar relationship between frictional resistance, and yarn sett. He concluded, "the frictional resistance of plain weave fabrics is sensitive to small changes in yarn geometry produced by altering yarn crimp, thread spacing, crown height, and fabric balance" (p. 91).

Fabric Hand, Including Stiffness and Surface Roughness
Fabric hand, surface roughness, drape, and stiffuess may all play roles in the damage done by fabric-to-fabric abrasion. The evaluation of the sensory properties of hand and drape in a textile is challenging. Subjective testing is dependent on the skill of the evaluators. To effectively evaluate hand and drape, each property must be broken down into its component tactile characteristics such as stiffness, roughness, fuzziness, springiness, and mechanical characteristics such as force to compress, tensile strength, tensile stretch, etc. (Collier and Epps 1999). Quantitative test methods are available for objective evaluation of a few of the tactile characteristics.
A cantilever bending test measures flexural rigidity, which is a measure of the relationship of stiffuess to fabric weight (Pierce 1930). Kadolph's (1998) definition of fabric stiffuess is "a measure of a fabric's resistance to bending or flexing" (p. 213 ).
In physical terms, more force is required to bend a stiff fabric than a limp one.

41
Kawabata suggested that quantifying the hand of a fabric required the evaluation of six mechanical and physical properties. These included: tensile properties, bending properties, surface properties, shearing properties, compressional properties, weight and thickness (Kawabata 1980 Development of the KES system allowed researchers to "relate objective measurement of the important properties in fabric hand to subjective evaluation" (Collier and Epps 1999 p. 269). The instrument was developed for use on fabrics of different construction, weights, weaves, and composition (Sabia and Pagliughi 1987). Chen (1992) noted "a particular need to quantify the relationship between different knit constructions and hand properties," (p. 200) and the KES is ideal for doing that research. Today, the KES is used primarily in the development of new fabrics and the evaluation of fabric finishes (Collier and Epps 1999).
Surface roughness may have an effect on several other textile properties. For example, in research on friction in yam spinning, Shick ( 1977) found that "an increase in surface roughness resulted in an increase in charge generation and is more pronounced when the yam passes a rough guide surface than when passing a smooth one" (p. 52). In addition, Harlock (1989) stated that the mechanical property of surface roughness in fabric relates to the quality and mechanical performance characteristics of abrasion and pilling resistance, but not to abrasiveness.

42
Temperature and humidity changes can cause subtle changes in fabric dimensions, including length. Gravity exerts a steady influence on textiles hung vertically. These and other forces on a fabric create strain, which is the change in length of a stretched fabric divided by the original length; this also is called elongation and stretch. Stress is the force causing this strain. Growth is defined as the permanent increase in length of a fabric after application of a load of a fabric. Elongation, growth, and growth recovery can act as indicators of the stability of a fabric (Collier and Epps, 1999;Tortora and Merkel 1996). Textiles stretched to less than breaking point do not immediately recover to their original dimensions. The elastic recovery is dependent on the force used, the length of time the force was applied, and the time allowed for recovery (Saville 1999).
Textiles in storage and display conditions seldom encounter breaking force stresses, \.
but the subtle long-term changes caused by gravity and humidity may affect elongation and elastic recovery. This, in turn, may affect the effectiveness of an overlay.
Electrostatic charge can attract and hold particles such as dust particles and chemical pollutants onto the surface of a textile. These in tum can become third-party abradants when one textile moves against another textile (Pye 2001;Collier and Epps 1999). Kadol ph (1998) defined electrostatic propensity as "a measure of the capacity of a non-conducting material to acquire and hold an electrical charge through friction or other means." She defined electrostatic cling as "the propensity of one material to adhere to another because of an electrical charge on one or both surfaces" (pp. . Electrostatic conductivity is the propensity of a fiber to carry or transfer electrical charges. Fabrics with low conductivity build up electrical charges. Poor conductivity is related to low moisture regain, and synthetic fibers tend to have lower moisture regain than natural fibers and thus lower conductivity and more problems with electrostatic charges (Collier and Tortora 2001). Finishes may be applied to fabrics to improve surface moisture retention and raise conductivity. These finishes help to reduce the problems caused by electrostatic cling, but the stability of the chemical composition of the finish over long term storage or display is a concern. 44 \. done by this researcher indicated that conservators used a variety of fabrics as sheer overlays. The six fabrics listed in the survey (silk crepeline, Stabiltex, bobbinet, tulle, nylon net, georgette) were selected for study. The survey did not specify the fiber content of tulle or bobbinet. Tulle was found in the marketplace in cotton, silk, and nylon, so all three were included. Fabric retailers labeled bobbinet as English net, and it was available in both nylon and polyester, so both of these were purchased. Nylon illusion, one of the fabrics listed under "other" in the survey, was added to the list.

Survey
One of the retail fabric stores carried a net labeled as polyester net, with a different knit structure than the other fabrics, so it also was selected. In total, eleven fabrics were tested. See Figures 1 to 11 for digitally scanned images of the fabrics with warp direction laying vertically. Fabrics were purchased from a local retail fabric store, a fabric wholesaler, two mail order fabric retailers, and from a conservation supply house. See Table 1. The cost of fabrics ranged from $70.00 per yard for silk tulle to $.69 per yard for nylon net. New sheer fabrics will appear in the marketplace as fashions change and new textiles are   developed. These fabrics will need to be evaluated for appropriateness before being used as overlays in conservation.

50
Because the fabrics in this study were purchased from a variety of retail and wholesale sources, and not all of the retail buyers were willing to give out their sources of manufacture or knew the sources, contacting the manufacturers of all fabrics was not possible. The manufacturer of three of the fabrics (nylon net, nylon illusion, and nylon tulle) stated that he applied a "chemical heat set finish to create a firm hand," but he did not provide further details (Ed Falk, telephone interview, 26 Feb 2004). Information about finishes including the chemical make-up of the finishes and method of application is unknown for the other eight fabrics.
All structural levels, fiber, yam and fabric, of the textiles were considered during this research. "Fabric is an extremely complex structure for mechanistic analysis. There exist several structural levels from fibers to yams and eventually to the fabric. Each level has its own geometrical and mechanical variables which control or influence to varying degrees the fabric behavior" (Pan 1996, p 312). Figures 12 to 22 show images of each fabric magnified at 5x except for Figure 17, which is magnified at 2.5x to have an entire hexagon visible in the picture. Three of the fabrics are woven in a plain weave (Figures 12,15,and 19). The warp knit nets are produced using part-threaded guide bars and altering the overlaps (Figures 13,14,16,17,18,20,21 and 22). In normal warp knitting, every needle must receive at least one overlapped thread, but nets can be produced because the same guide bar does not have to supply every needle, nor does every needle need to be overlapped by the same number of ..
Illusion, nylon (5x) Figure 17. Net,nylon (2.5x) ...  Two are hexagonal nets; one is a diamond net; and all three have different knit structures, yet all are called tulle. Even consulting the textile experts does not resolve the confusion. Sometimes fiber content is specified in a definition, sometimes weave or knit structure, and sometimes end use. For example, crepeline is defined by Landi as a "fine, plain weave fabric of silk or polyester," by Picken as a ''thin lightweight dress fabric of silk or silk mixture," and by Totora as "an exceptionally sheer, plain weave, silk fabric similar to chiffon" (Tortora and Merkel 1996 p. 149;Landi 1998 p. 198;Picken 1985 p. 87). Tulle is defined by Fairchild's Dictionary of Textiles as "a net with hexagonal mesh made on a warp knitting machine of silk, cotton or manufactured fiber" (Tortora and Merkel 1996 p. 592). Illusion is a "fine, sheer net fabric" and nylon net is a "sheer net made of nylon" (Picken 1985 p. 182). Figure 24 illustrates a net structure similar to illusion (Fig. 5) and the fabric labeled as nylon tulle ( Fig. 10).
The fabric labeled as polyester net ( Figure 18) has a structure similar to sandfly net, which is illustrated in Figure 25. Bobbinet, the fabric name used in the survey, is defined by Fairchild's Dictionary of Textiles as a machine-made net with almost hexagonal meshes of twisted cotton or silk yam, and English net was defined as "a net with hexagonal meshes" (Tortora and Merkel 1996 p. 61, 201). Two of the hexagonal mesh nets used in this research were labeled as English nets by their retailers.
Georgette is defined as "a sheer, lightweight, plain weave silk or manufactured fiber fabric" (Tortora and Merkel 1996 p. 243). Stabiltex is a trade name for a plain woven polyester fabric imported from Switzerland, and its American distributor describes it as "stronger and longer lasting than crepeline." It also is called Terelene and Tetex (Talas, undated p. 73). Care must be taken to avoid confusion when discussing and comparing commercially named fabrics. Retail names of the fabrics plus fiber content are used throughout this research, but identifying them by knit structure and fiber content would be more precise. The fabrics were purchased in the spring of 2003. They were tested as received from the retail and wholesale sources; they were not washed or pre-treated.
A 100% red cotton flannel was chosen as a fiber donor fabric for the abrasiveness testing based on research done previously by Simpson (1991Simpson ( , 1993. The retailer called it a D/Nap flannel, with a fabric count of 44 x 44 yams per inch, and it did not have a flame resistant finish. The flannel was washed with ionic and nonionic surfactants in warm water in a standard washing machine and dried in a home dryer prior to being used in the abrasion test.

Yarn Characteristics
A variety of fiber and yam types make up the fabrics chosen for this research.
The yarn characteristics affected the performance of the fabrics and were useful in the analysis. See Table 2 for yarn characteristics. One fabric was made of a cotton yarn with staple-length fibers; two fabrics were of silk multi-filament yarn; three were of polyester multi-filament yarn; one was polyester mono-filament; two were nylon mono-filament and two were nylon multi-filament. Of all the synthetic fibers, only the nylon in the nylon net fabric did not have delustrant added.
Yarn denier, a measurement oflinear density, was obtained from the retailer or the manufacturer when possible. When it was not available denier for the woven fabrics was calculated by weighing three one meter long yarn samples in both the warp and filling directions to 0.0001 gm and then multiplying the mean of those samples by 9,000. Obtaining one meter samples from the weft direction of silk crepeline or the knitted nets was not possible, but shorter yam samples were unraveled, measured, and weighed. Weights were scaled up to the weight of one meter and -..l denier was calculated by multiplying by 9,000. Because the samples were very small, accuracy was compromised.
The deniers from the retailers and manufacturers were for yams prior to the application of any finishing. The calculated deniers are from fabrics that may have applied finishes, and a finish could affect the weight of the yams and affect accuracy.
Because of these discrepancies, denier is reported in Table 2, but it is not used in any of the analyses. Normally cotton count would be reported for a cotton yam, but to facilitate comparisons in this research, the cotton yarn linear density also is reported as denier.
Other yam characteristics such as luster, yam type, spin, and yam cross-section were determined using microscopes located in the Textiles, Fashion Merchandising and Design Department at the University of Rhode Island. Image-Pro Plus, version 4.5 for Windows, by Media Cybernetics, Inc. was used to measure yam and filament diameter.

Fabric Structure
The fabric structure of the sheer overlays affected performance test results.
Three of the fabrics-silk crepeline, polyester georgette, and polyester Stabiltex-are woven fabrics with a plain weave structure. The other eight fabrics are warp knits.
Thomas stated that the physical properties of a warp knit are a function of its structure.
In this, it differs from weft knits and woven fabrics (fhomas 1976). Three were knit on tricot machines; the other five were knit on raschel machines.
Either tricot or raschel machines can produce knit meshes, depending on the complexity, yam size and end use (Kadolph and Langford 1998). Falk Industries, the manufacturer of the nylon tulle, described it as a tricot fabric. The buyer at Baer fabrics described the polyester English net as a tricot fabric. The polyester net, from an unknown manufacturer, unraveled similarly to a tricot knit under close examination. See Figure 25 for an illustration of its structure. The complex structure of the other five fabrics defines them as raschel knits. Nylon illusion, polyester net, and nylon tulle are diamond mesh nets; the polyester net is similar to a net described by Spencer (1989) as a sand-fly net. Cotton tulle, silk tulle, nylon net, and the two English nets are hexagonal mesh nets. The Tortora and Merkel (1996) definition of tulle as a net with a hexagonal mesh, would label the two English nets, seen in Figures   13 and 14, as a tulle structure, despite their marketing names (p. 592). The nylon tulle, due to its diamond mesh seen in Figure 21, would not fit this definition of tulle. The nylon net does not have the knit pillar structure of a typical bobbinet, but the cotton and silk tulles and the English nets both have the pillar structure (see Figures 13,14,17,22). The relative stability or stretch of a warp knit fabric can be altered by control of the knitting stitch (Kadolph and Langford 1998). The diamond mesh nets in Figures   16 and 21 have different yam interlacement and stretch properties than the hexagonal mesh nets in Figures 13,14,17,20,and 22. The illustrations in Figures 23 -25 show these differences clearly. The polyester net, which is similar to the sand-fly net pictured in Figure 25, may have unique properties due to its yarn arrangement.
Finishes may have been applied to the fabrics, but because they were obtained primarily at retail establishments, full information on finishes is not available.
According to the manufacturer, three of the fabrics-nylon net, nylon illusion, and nylon tulle-do have a "chemical, heat set finish to create a firm hand" (Ed Falk, telephone interview, 26 Feb 2004). According to Spencer, "there is considerable potential for changing the fabric properties during the finishing process as well as during knitting" (Spencer 1989 p. 40). A manufacturer's finish can change the hand 60 of warp knits; it can add stiffness or softness a fabric. Often anti-static agents may be included in the finishing process (Thomas 1976). This lack of knowledge about fabric 1 finishes prevents analysis of their effect on various properties.

Standard Testing
A sampling plan was created, and test samples were cut using a rotary cutter and mat. Fabrics were conditioned in the conditioning room at 70°±2° and 65%±2% relative humidity for 24 hours prior to each test according to ASTM D 1776-98 (ASTM 2003. Testing was randomly assigned to samples. All standard testing was done in the Textile Performance Laboratory at the Department of Textiles, Fashion Merchandising and Design at the University of Rhode Island.
Standard tests from ASTM, International were used for basic textile descriptive tests such as thickness, weight, and fabric count. Standard tests were used whenever available to test the fabric properties important to conservators who use sheer fabrics.
Non-standard tests were employed when standard tests did not exist to measure other

Cover
Cover is "the ratio of fabric surface occupied by yarn to the total fabric surface." It is sometimes called the cover factor. Cover may be calculated for woven fabrics when warp and filling yam diameters and fabric count are known (Kaswell 1963 p. 450). The structure of knitted nets makes a straightforward geometrical calculation of cover very difficult. Digital imaging software was used in this research to provide a measure of cover that allowed comparison of cover factor between knitted and woven fabrics. Image-Pro Plus, version 4.5 for Windows, by Media Cybernetics, Inc. was used to analyze the percent area covered by each fabric (Media-Cybernetics 2002). Fabrics were scanned using a Hewlett Packard 1200 series scanner at 1200 dpi with dark purple Pantone paper # 19-3 714 as background. HP Director 7 .1. 4 was used to digitize and save the images.
A one-square-inch area of interest was defined at random on each scanned image. The Image-Pro software measured the percent area of white fabric for each area of interest. Three areas of interest were defined and measured for each fabric sample. Mean percent area was calculated for each fabric. This measure allowed the eleven fabrics to be compared to one another for area covered or conversely for the amount of background that shows. This measure was not compared to calculated cover factor for individual woven fabrics, but in a situation where fabrics are ranked most cover to least cover, it should provide a similar ranking. Digital imaging makes the measurement of cover for complex fabric structures such as knits possible. This method of measuring cover is easier to do and may prove to be a better method for comparisons than the calculated method.

Gloss
Gloss measurements were performed by Bossa Nova Tech, a company specializing in non-destructive testing instruments, located in Venice, California.
Bossa Nova Tech used the Samba Advanced Vision System. The Samba Live Gloss Measurement software calculates "the average gloss degree in a Region of Interest." The Samba system uses an Analog PAL sensor video format; with a 9-hertz refresh rate, a resolution of 768 x 576 and one half-inch CCD type. The spectral bandwidth is 400-700 nm, and the gloss degree measurement ranges from 0 -100%, with a degree of accuracy within 0.5%, and a scattering contrast ratio of 100 (Breugnot 2003).
Results were reported in gloss degree percentage. Three areas on each fabric sample were tested, but results were received as single values, so no statistical testing was done as replicate test scores were not available.

Coefficient of Friction and Surface Roughness
Two tests, surface roughness and coefficient of friction, were measured on the Kawabata Evaluation System (KES) to evaluate hand. The system combines tests for tensile, shearing, pure bending, compression, and surface roughness characteristics. It also tests the coefficient of friction on fabric surfaces (KatoTech 2004). The Kawabata test is not recognized as a standard test by AA TCC or ASTM but is used in textile research when quantitative data on the handle of fabric is required (Kawabata 1980;Collier and Epps 1999). Friction is the ability of a fabric to move along another surface and may be affected by surface roughness and fabric structure. Surface roughness is defined as the "deviation in thickness on the surface as a result of the structure" of a fabric (Scruggs 2004  The coefficient of friction (MIU) is a value between 0 and 1 and indicates the amount of resistance/drag sensed by the probe as it moves across the fabric surface. A higher value indicates greater friction. This result is largely associated with amount of contact area the probe makes with the fabric surface. The greater the contact, the higher the MIU. Therefore, this result is sometimes inconsistent with expectations. For example, very soft fabrics may feel smooth and slippery between the fingers when you feel them, but you may get a higher MIU than with a stiff fabric because the probe may press into a soft fabric, almost becomes imbedded, and result is a high contact area/higher MIU. Stiff, rough fabrics may have lower MIU than expected because stiffness prevents the probe from making total contact with the fabric surface. (Scruggs 2004) Many of the fabrics tested in this research were stiff, and the nets had large spaces between yarns that may have increased the surf ace roughness and changed the contact area of the probe. Results were interpreted considering this caution.

Abrasiveness
Most of the large body of research on abrasion in textiles investigated abrasion resistance, not abrasiveness. Seven standard tests measure abrasion resistance in fabrics using various mechanical means to abrade the textile to a breaking point. None of these tests uses fabric-to-fabric abrasion, and all of them use a high force to abrade the test fabric to a hole or break. The following crocking and frosting tests also simulate abrasion: AA TCC 8-1996 (Collier and Epps 1999). Simpson (1991Simpson ( , 1993 used a crockmeter and a variation of the standard crocking test to simulate low-force fabric-to-fabric abrasion such as might occur in museum display or storage conditions. Annis went a step further and created a testing machine to simulate low-force fabric-to-fabric abrasion. See Figure   cotton undyed print cloth. A fiber-donor fabric, red 100% cotton flannel similar to that used by Simpson in her research, was placed on the upper pad (Simpson 1991(Simpson , 1993.
After pre-testing several combinations of load, speed, cycles, and orientation, this researcher chose 40 rpm, 800 ± 30 g load, and 50 cycles of orbital abrasion. While lower speed, load, cycles, and linear direction would better simulate actual abrasion in display and storage settings, they did not show enough released fibers for adequate digital image analysis. These conditions are still low compared to those used in the standard test method tests for abrasion resistance. Three replications were done for each test fabric. Test samples were packed carefully for transportation back to the University of Rhode Island for digital image analysis.

69
Three, randomly selected, 2-inch square areas of each test fabric were scanned on an Epson Perfection 1200 Photo scanner at 1200 dpi. IrfanView 3.85 was used to digitize and save the images. Thus nine images of each fabric were available for analysis. Image-Pro Plus, version 4.5 for Windows, made by Media-Cybernetics, Inc. was used to analyze the area of red cotton flannel fibers left on the test fabrics after abrasion on the ABD Materials Evaluator (Media- Cybernetics 2002). The count/size function was used to highlight each red fiber in the image. Addition of various filters seemed to increase the likelihood that shadows would be counted as fibers. The data used were from unfiltered images. Data were reported as area (mm 2 ) of red fiber left on the overlay fabric after abrasion.
The digital imaging software also counted the number of fibers, fiber fragments, and fiber clumps as objects, but some samples had large clumps of fiber, while others had only small individual fibers and fiber fragments. The software could not distinguish objects by size. Control of fragment size is necessary before count will accurately reflect the amount of fiber left on the overlay after abrasion. Count data were not used in the statistical analysis but are shown graphically in the results section.

Statistical Analysis
The null hypothesis of the research was that overlay fabrics do not differ in abrasiveness. Raw data were recorded in an Excel spreadsheet and input  One-way analysis of variance (ANOVA) was performed on the overall mean weights of the eleven fabrics to determine the existence of significant differences between at least two fabrics. The ANOV A showed significant differences existed between at least two fabrics (p::;0.01) in the weights of the fabrics, therefore Tukey homogeneous subset (HSD) analysis was done. Tukey's HSD method produced seven statistically significantly different groups (not necessarily mutually exclusive) from the eleven fabrics.     Figure 28. Thickness of overlay fabrics A one-way ANOVA was performed on the overall mean thicknesses of the eleven fabrics to determine the existence of significant differences between at least two fabrics, and it showed that significant differences existed at the (p<0.01).
Therefore, the data were analyzed by Tukey's HSD method. This analysis method produced six statistically significantly different subsets (not necessarily mutually exclusive) from the eleven fabrics as shown in Table 4  the one used here was labeled "super-fine." Figure 29 shows the hex per inch measurements graphically.    Table 2 for yarn denier data.    Table 7.

Cover
As measured by digital imaging software, woven polyester georgette, with its high fabric count, had the highest percent cover at 54.26% and was, therefore, the least sheer. See Figure 31. The cotton tulle, with its fuzzy staple-length fibers partially covering the interstices, and a high hex-per-inch count was the second least sheer at A one-way ANOVA comparing the means of all eleven fabrics indicated a significant difference between at least two fabrics (p< 0.01 ). Tukey HSD analysis of cover produced nine homogeneous subsets. See Table 8. It indicated that the four least sheer fabrics, those with highest cover, (the two English nets, cotton tulle, and polyester georgette) were significantly different from all other fabrics and from each other. The sheerest fabric, nylon net, also was significantly different from all other

86
Positive correlations between cover with thickness and cover with weight are also an indication that thicker, heavier fabrics are likely to be less transparent than thinner, lighter fabrics. The negative correlation between cover and electrostatic cling corroborates the correlations seen between fabric count and cling: the high cover, high fabric count wovens have lower cling times than the loosely knit, low cover, low hex count knitted nets.
Cover also can be used as an indicator of how much protection an overlay fabric offers from ultraviolet radiation and airborne particulate soils. The more open sheer fabrics with greater spaces between yarns allow light and particulate matter to affect the historic textile more than less sheer fabrics with higher cover.

Gloss
The Samba Advanced Vision System at Bossa Nova Tech, in Venice, California measured gloss. Stabiltex, made of the finest multi-filament yarn, was the glossiest fabric by a large margin at 26.43%. See Figure 32. Silk crepeline, another woven fabric, was next most glossy at 15 .21 %. This finding reflects the opinions of conservators such as Blum and colleagues that Stabiltex and silk crepeline give objects a sheen (Blum, et al. 2000). The two English nets, nylon net and polyester georgette, were clustered between 11 and 12%, while silk tulle, nylon tulle, polyester net, and nylon illusion clustered between 7.5 and 8.5%. Cotton tulle was the Yarn twist in multi-filament yarns is important in determining the amount of gloss according to Kim and Shin (2004). They noted that as twist in multi-filament Yams increases, the luster unit size diminishes, which results in a macro-level gloss decrease. Stabiltex, the glossiest fabric, was woven of large diameter filaments (44 µ) that were only slightly twisted, producing large size luster units and high gloss which 18 consistent with Kim and Shin's research. Yarn twist was not quantified in this research because of the difficulty of unraveling a piece from the knitted nets long enough to evaluate. Testing the yarns prior to net manufacture for yarn twist and gloss degree could provide that data. Statistical analysis was not done on degree of gloss measurements because the data from Bossa Nova Tech was presented as single points for each fabric and multiple trial data were not available.

Abrasiveness
The measure/count capability of digital image analysis software was used to quantify the effects of the abrasive action of sheer overlay fabrics using the ABD Materials Evaluator. Figures 33 and 34 show examples of the red cotton flannel fiber, fiber fragments, and fiber clumps visible on the overlay fabrics post-abrasion. The software calculated the area of red cotton fiber deposited on the sheer overlay fabrics by abrasion. These data are indicators of the abrasiveness of the fabrics tested. Table 9 shows the area of cotton flannel fiber on the sheer overlays after abrasion. Nylon net loosened and transferred the most fiber from the cotton flannel fabric. The three tulle fabrics had the next three highest amounts of fiber transferred to their surface after abrasion. Polyester georgette was the least abrasive fabric,   A one-way ANOVA showed a significant difference between at least two fabrics for the area of fiber deposited by abrasion (p<0.01). Tukey's HSD method, based on mean abrasiveness, produced two statistically significantly different groups (not necessarily mutually exclusive) for abrasiveness as measured by area of fiber left on the overlay fabric. See Table 10. These groups were highly overlapping homogeneous subsets with high degrees of separation at p=0.067 and p=0.064 respectively. The analysis showed that nylon net was significantly more abrasive than polyester georgette and polyester English net.
The digital imaging software also produced data by counting the number of objects. The software could not distinguish objects by size; therefore, large clumps of fiber, small individual fibers, and fiber fragments were each counted as one object.
This technique needs modification to account for object size. Statistical analysis was not done on these data, but they are presented graphically along with the area of fiber in Figure 36. This graph shows that the three woven fabrics and polyester English net had the lowest number of fibers, fiber fragments, and fiber clumps present after abrasion. The fabrics with the highest object counts after abrasion were all knits.
Nylon net had the highest count, again indicating that it is the most abrasive. The one fabric made from a staple fiber, the cotton tulle, was at the high end of the range because the protruding ends of the cotton fiber in the tulle could easily entangle the    The positive correlations with warp and filling growth indicate that factors that affect fabric growth, such as fabric structure, also may be variables of interest in abrasiveness results. Fabric weight and thickness were not significantly correlated with abrasiveness. These findings are consistent with Simpson's (1993) research in that fabric construction had more affect on the abrasiveness of backing fabrics than did weight and thickness.
Cover and fabric count both correlate negatively with abrasiveness. Cover and fabric count correlate positively with each other. The woven fabrics in the study had high fabric counts and high cover but low abrasiveness. The knitted nets had lower cover and lower fabric count with higher abrasiveness results. Determining fabric count is easily done, and cover can be estimated by eye without a detailed measurement, so these two properties may be useful as a quick estimate of abrasiveness.

97
The correlation with stretch and growth is similarly easy to use when choosing an overlay. Conservators can estimate the amount of stretch in an overlay fabric without the need for expensive testing equipment. Based on data in this research, the least abrasive overlay fabric would be one with low stretch, high fabric count, and high cover cover. Unfortunately, high fabric count and high cover mean that a fabric is less transparent. Cover, fabric count, and stretch may be used as predictor properties for abrasiveness to assist conservators in choosing overlay fabrics that will not be abrasive.

Coefficient of Friction
The Kawabata Evaluation System measured the coefficient of friction. Data were reported as means in both warp and weft directions and as a grand average, which combined data from both directions into one value. One-way ANOV A was performed on the overall mean friction coefficients of the eleven fabrics to determine the existence of significant differences between at least two fabrics. The ANOV A proved significant differences existed between at least two fabrics (p.::;O. 01) in the friction of the fabrics, therefore Tukey homogeneous subset analysis was done. An ANOVA could not be performed on the grand average data because the KES produces only one value per fabric, and replicated values were not available for analysis.
Polyester net had the lowest coefficient of friction in the warp direction. See Table 12. Cotton tulle and silk crepeline also had friction coefficients at the low end Polyester net had the lowest coefficients of friction in the filling direction as it did in the warp direction. See Table 13. Polyester Stabiltex and silk tulle had the highest filling friction coefficients. Three of the four nylon fabrics still had high friction in the filling direction, but the order within those four fabrics was reversed: English net had the highest, then nylon tulle, and nylon net had the lowest. Nylon tulle had a low ranking for filling friction. Neither fiber content nor fabric structure were grouped in the filling direction as they were in the warp direction.
The differences between rankings reported in Tables 12 and 13 probably are determined by the structures of the knits. Fabric finishes also can affect the measurement of the coefficient of friction, but because information about finishes on the tested fabrics is incomplete, the impact of the finishes cannot be analyzed. These coefficient of friction results are graphically displayed in Figure 39.    The KES grand average for each fabric that combines the data from the warp and filling directions is in Figure 40. The four nylon nets continue to be at the high end of the range for overall friction, and the cotton tulle is near the low end. Collier and Epps ( 1999) reported that ny Ion fiber has a low coefficient of friction, yet in the warp direction in this test they are high possibly because fabric structure is the major contributor.
A higher coefficient of friction means that more force is required for these fabrics to slide across another surface. In conservation, materials sometimes are chosen because their surface has a higher friction and is able to grab or hold onto another fabric. This is important when choosing a support fabric for a textile displayed on a slanting mount board with little or no stitching to hold it in place.
Overlays do not need to support a fabric the way a backing fabric does, but a fabric Statistical correlation coefficients were calculated between coefficients of friction and selected variables from the data set using Spearman' s rho method. Warp coefficient of friction was negatively correlated with fabric count in the warp direction (r=-0.726; p=0.01) and cover (r=-374; p=0.01). This means that fabrics with high cover and high fabric count had low friction. An example would be the cotton tulle, which has high cover because the staple fiber ends protrude from the yams increasing the surface area of the fabric, the cover and the coefficient of friction. Warp friction was positively correlated with stretch (r=0.587; p=0.01) and growth (r=0.590; p=0.01) in the filling direction, but not in the warp direction. Filling friction was not correlated with stretch or growth in the warp or filling directions.
Neither warp nor filling friction correlated with the abrasion data. Friction was not correlated with electrostatic cling data. Only warp friction was correlated with warp roughness (r=0.631; p=O.O 1 ); filling friction was not correlated with warp or filling roughness, and warp friction was not correlated with filling roughness. This is a surprising result, as one would expect that friction and roughness might be correlated, especially in light of Ajayi's (1992)  Examining fabric properties with respect to surface roughness showed that the two raschel knit English net fabrics had the highest surface roughness in the warp direction. See Table 16. The three woven fabrics had low roughness, but a tricot net (polyester net) also ranked among the lowest. Fiber content, yam structure, and knit structure do not fall into any logical groupings within the warp surface roughness measures. Within the filling surface roughness rankings, fabric structure clearly divides the fabrics into groups. All the raschel knit fabrics are above the median of the roughness scale in the filling direction. See Table 17     controlling for warp and weft directions of the overlay fabrics, is needed to provide additional data with which to explain these results.

Elongation
Results from elongation tests are in Table 21. The data were measured in millimeters but are expressed here as a percentage of total. The data set was incomplete because both samples of nylon net broke before the end of the test in the filling direction, and one sample of the nylon illusion broke in the warp direction.
Nylon illusion had the greatest stretch in the warp direction almost doubling in length.
It, along with the next four fabrics having the greatest amount of stretch in the warp direction, also had the greatest amount of growth in that direction. The illusion and the two others that rank high in stretch and growth are tricot knit fabrics made of mono-filament yarns.   Table 21 are higher for the filling direction than the warp direction. Nylon tulle had a higher stretch in the filling direction than all but one fabric, and all three tricot fabrics were above the median for filling stretch. The knitting technique of skipping underlaps necessary to create an open net increases the stretch of the fabric. An applied finish can affect the stretch and growth of a fabric. As expected, the three woven fabrics exhibited the most warp and filling stability with very little stretch (:S 3.0%) or growth (:Sl.0%).
One-way ANOV A was performed on the overall means of the eleven fabrics to determine the existence of significant differences between at least two fabrics. The ANOVA showed significant differences existed at the (p<0.01) in both the stretch and growth of the fabrics, therefore Tukey's Homogeneous Subset Analysis was done for both stretch and growth.
Tukey' s HSD method produced eight statistically significantly different groups (not necessarily mutually exclusive) for stretch in the warp direction. See Table 22.
Nylon illusion, silk tulle, and polyester net were each significantly different from all the other fabrics and exhibited the highest warp-wise stretch. Nylon English net also was significantly different from all the other fabrics but fell near the low end of the scale for warp-direction stretch. The three woven fabrics had the least stretch in the warp direction and were not significantly different from each other, but the set of three were significantly different from all of the knitted nets.
Tukey's HSD method produced four statistically significantly different groups (not necessarily mutually exclusive) for stretch in the filling direction. See Table 23.
In the filling direction, only nylon net was significantly different from all of the other Fabrics, and it was at the high end of the range. The wovens had the least stretch in the filling direction because the plain weave fabric structure gives them little stretch.
Cotton tulle joined the three wovens in the lowest subset.       These results also may be due to fabric structure. Because technical back and face were not controlled during abrasion, fabric face might be another confounding variable. Controlling all these variables could help to clarify these results.
Stretch and growth also positively correlated with electrostatic cling, and negatively with fabric count. The high count woven fabrics had the lowest amount of stretch and growth, while the low count, knitted nets had higher stretch and growth.
This makes sense when considering fabric structure, as knitted nets are designed to have stretch.

Electrostatic Cling
Electrostatic cling data are presented in Figure 47. Four of the sheer overlays clung to the metal plate for the maximum time of I 0 minutes in the filling direction.
These were polyester net, silk tulle, nylon tulle, and nylon illusion. Three of these, excluding silk, are tricot fabrics made of mono-filament yarn. See Table 27. Both polyester net and silk tulle also showed maximum cling time in the warp direction.
Nylon tulle (9.21 minutes) and nylon illusion (  . ---   (4.39 minutes) cling in the filling direction. The two English nets and the nylon net fell in the middle of the range with cling times ranging from 2.76 minutes to 8.36 minutes.
The woven fabrics had very low charge build-up leading to static cling. Also, as expected, the cotton fabric had little cling. The three fabrics made of monofilament yarns had the highest charge build up indicating that their high static cling also would make the fabrics attract charged airborne particles. These airborne particles could act as third-party abradants or chemically damaging pollutants when brought into proximity with the fragile textile under the overlay.
One-way ANOVA was performed on the overall means of the eleven fabrics to determine the existence of significant differences between at least two fabrics. The    using Spearman's rho method, it was positively correlated with stretch, growth, roughness, and negatively correlated with fabric count and cover. The woven fabrics with high fabric counts and high cover had very low cling times. The cotton tulle fabric also had high cover, even though it was a knit and these factors, combined with the low conductivity of the fiber, made it fall in the same range as the wovens. The knitted nets with lower fabric counts and lower cover had higher cling times. Yam and fabric structure, and fabric finish are the major variables affecting these results.

Stiffness
The data indicated that ny Ion net was the stiffest fabric when measured in the warp direction, while polyester net was the stiffest in the filling direction. Polyester georgette and the English nets were the least stiff in both directions. See Figure 48.
Not all of the fabrics had higher stiffness ratings in the warp direction than the filling.
One-way ANOVA comparing the means for all eleven fabrics indicated that .   tricot mono significant differences in stiffness exist between at least two fabrics (p<0.01).
Tukey HSD analysis produced five homogeneous subsets (not necessarily mutually exclusive) in the warp direction. See Table 31. Nylon net, the stiffest fabric, was significantly different from all other fabrics in the warp direction. Polyester georgette and polyester English net, the least stiff, were significantly different from all other fabrics in the warp direction but not from each other. The other three subsets in the warp direction were highly overlapping and had high degrees of mean separation within their subsets.
Tukey HSD analysis produced four subsets in the filling direction. See Table   32. Polyester English net and polyester georgette, the least stiff in the filling direction, were not significantly different from each other. In the second subset, polyester georgette was not significantly different from nylon English net, nylon illusion, and nylon tulle -one raschel and two tricot knits. The three large subsets in the filling direction were highly overlapping and had high degrees of mean separation within their subsets.
Statistical correlation coefficients were calculated using Spearman's rho for the stiffness data paired with other variables. Stiffness did not correlate with abrasiveness or any other variables except thickness and weight. Stiffness correlated negatively with thickness and weight. See Table 33. The confounding variable of finish might be a factor in these results as a finish, which imparts stiffness might also add weight and thickness to a fabric. The stiffness test results create more questions than they resolve. What happens to the fabric stiffness if it is washed before being used? When a conservator desires some stiffness to stabilize a fragile textile, is it safe to leave an unknown chemical finish in the overlay fabric and hope that it is stable in the long term? The knowledge that fabrics have different stiffness in the warp or filling direction can be put to good use in conservation. Overlay fabric can be applied with its stiffest direction corresponding to the stiffest direction of the fragile fabric to re-create the original hand or drape of the fabric or it may be applied with the stiffest direction of the overlay along the weakest direction of the fragile fabric to provide some stability.
For example, an overlay on a skirt, which will be exhibited on a mannequin, should have the stiffness of the overlay correspond with the stiffness and hand of the original skirt fabric so as not to change the drape of the costume. A flat textile with a loss of yams and loss of structural integrity in the warp direction would benefit from having an overlay with stiffness applied in the warp direction to supply stability.
Conservators should consider both the stiffness of an overlay fabric and also the direction of the stiffness and use that knowledge to enhance treatment outcomes. The photomicrographs on pages 48 and 49 can be used in conjunction with this table to identify the important structural characteristics of a specific fabric. Using a binocular microscope or a high-powered magnifying glass, a conservator can determine the shape and size of the interstices in any sheer fabric. The interlacement pattern of the yams can be compared to the photographs in this research, &nd then the performance characteristics of that type of fabric can be looked up on Table 34. This summary of information has not been available before to help conservators make an informed choice.   (Schick 1977). When selecting a fabric for a given purpose, a person must "assess many properties simultaneously and subjectively and rank the fabrics in order of preference" (Booth 1969 p. 282). This research assessed the properties of sheer overlay fabrics to assist conservators in matching the characteristics of the textiles to the needs of the fragile object being conserved.

Overall Performance Results
The primary objective of this research was to compare the abrasiveness of sheer overlay fabrics and to identify predictor properties for abrasiveness. This research also analyzed and compared performance properties of the sheer textiles to provide conservators with objective data. Statistical analysis showed significant differences between fabrics and significant correlations between properties.
Analysis of variance found a significant difference between at least two of the fabrics tested for abrasiveness. Nylon net was the most abrasive fabric; polyester georgette was the least abrasive fabric. The three woven fabrics tested were all at the low end of the abrasiveness scale. The cotton tulle, with yarns of staple-length fiber in a raschel knit, ranked high in abrasiveness. The silk tulle and nylon tulle, a raschel and a tricot knit, also ranked high in abrasiveness. The type of filament yarn, monofilament versus multi-filament, did not affect abrasiveness, nor did raschel versus tricot knit structure. Hexagonal versus diamond shaped interstices in the knitted nets did not affect abrasiveness. Due to the method of testing abrasiveness, the differences between fabrics were small, but Tukey HSD analysis showed a significant difference in abrasiveness between the most abrasive fabric, nylon net, and the two least abrasive fabrics-polyester georgette and polyester English net. Conservators should not choose nylon net for use as an overlay because the abrasiveness of the overlay could damage the fragile or historic textile.
Predictor properties for abrasiveness were identified using Spearman' s correlation coefficients. Cover factor and fabric count both correlated negatively with abrasiveness; stretch and growth correlated positively. A fabric with low cover, low fabric count, and high stretch and growth is predicted to have high abrasiveness. The properties of cover, fabric count, and stretch/growth can be determined without sophisticated measuring instruments and can be assessed by conservators when choosing fabrics. The research did not show a relationship between abrasiveness and fabric friction, surface roughness, stiffness, or static cling. These findings are consistent with research by Harlock (1989) where he found no relationship between surface roughness and abrasiveness. In addition, no correlation existed between abrasiveness and thickness or weight. This lack of correlation is contradictory to results by Simpson (1993) who found that heavier and thicker fabrics were more abrasive. Simpson was testing backing fabrics that usually are heavier than overlay fabrics. All fabrics in this research were lightweight sheer textiles.
This research identified and quantified other properties of the sheer fabrics. Table 34 summarizes these properties. Results of the pre-study survey done by this author indicated that conservators consider sheerness to be the most important criterion in choosing overlay fabrics. This research ranked overlay fabrics for sheerness using cover factor as a measure. Fabrics also were ranked using fabric count, in yams per inch for wovens and hex per inch for knits. Both cover and fabric count can be used to assess sheerness. Nylon net had the lowest cover and fabric count; cotton tulle had the highest cover and fabric count of the knits. The three wovens had higher fabric counts than the knits, but polyester Stabiltex and silk crepeline ranked in the mid-range for cover. Polyester georgette, with the highest cover of all the fabrics, had the highest fabric count. One multi-filament knitted net with large interstices and three mono-filament knitted nets ranked the lowest for cover and could be chosen if sheerness were the only criterion.
Fiber content and color ranked as the next most important criteria for choosing overlay fabrics. These two factors can be assessed by conservators at the point of purchase. Yam characteristics and fabric structure contributed more to differences between sheer fabrics than did fiber content. Fiber content does affect the longevity of sheer fabrics; both silk and nylon deteriorate with exposure to sunlight (Collier and Tortora 2001).
The fourth ranked criterion of conservators was fabric hand. Hand is a very subjective characteristic, but some properties help interpret the feel and drape of a textile. Stiffness and surface roughness were used as objective measures in this research to assess hand. Surface roughness varied in the warp and filling direction, and this orientation may affect the hand of the fragile fabric in the completed treatment. Fabric structure, including type of knit, played an important role in surface roughness. The two English nets had the highest surface roughness in the warp direction; silk crepeline had the lowest. The three tulle fabrics had the highest surface roughness in the filling direction; polyester georgette and silk crepeline had the lowest.
Stiffness also was different in the warp and filling directions for several fabrics. Incomplete information about the finishes used on the fabrics in this research and their impact on the stiffness results indicates the need for further research in this area. In this research, nylon net was the stiffest fabric in the warp direction; polyester net was the stiffest in the filling direction. Polyester georgette and polyester English net were the least stiff in both warp and filling directions. Conservators should be aware of the differences in stiffness in warp and filling directions and use that knowledge to better match sheer overlays to an object' s needs.
Since both surface roughness and fabric stiffness affect the hand of a fabric, the addition of an overlay should be done to match the hand of the original fabric as closely as possible. On objects that will be framed or kept flat, the need to match the hand of the overlay and original fabric is not critical, but for a fabric that will drape such as a skirt or a set of draperies, the overlay should match the hand and drape of the object to preserve the original feel and look of the fabric. An overlay that will form to the surface of the object will be less noticeable than one which stands stiffly away from the surface of the fragile textile. An overlay that molds to the surface will do a better job at keeping small bits of damaged fabric in place than will one which stands away.
Elongation and elastic recovery, measured as stretch and growth, were assessed as indicators of strength and stability. The three woven fabrics-Stabiltex, silk crepeline, and polyester georgette-had the least stretch and growth of all the fabrics. Of the knits, nylon illusion had the most stretch in both warp and filling directions; polyester English net had the least stretch in both directions. Growth is a measure of how easily a fabric returns to its original length after a stretch with constant load and time. The nylon net broke before the end of the test in the filling direction and growth could not be tested. This breakage indicated that it is not a stable fabric under the stress of four pounds of load. In half of the tests, the nylon illusion broke in the warp direction. Despite the breakage, nylon illusion had the greatest growth in the warp direction; nylon tulle had the greatest in the filling direction. Of the knits, polyester English net had the least growth in the warp direction; cotton tulle had the least in the filling direction. Conservators should take the low strength of nylon net and nylon illusion into account when choosing appropriate overlays for a specific treatment. The fabrics that rated high for growth do not provide enough stability to function well as overlays that provide support for a fragile textile.
Thickness, weight, gloss, and static cling also were assessed and ranked in this research. Any of these properties could be important in a particular overlay application in conservation. These data are available in Table 34 to make selection of a sheer fabric more objective for the needs of each project.
This research also found that yarn structure is central to the gloss rating of a sheer fabric. The highest gloss fabrics were the fabrics made of large diameter multifilarnent yarns with little twist. The least glossy fabric was the cotton tulle, made of staple-length cotton fiber. Two of the mono-filament yarn fabrics, nylon illusion and tulle, also were low in the gloss ranking; the third mono-filament, nylon net, ranked higher because it had no added delustrant.
Using image analysis software to measure cover, as was done in this research, may prove to be easier and more accurate than the formula previously used to calculate cover. Certainly, the image analysis software allows cover to be determined for knits and non-woven fabrics in a way that was previously not possible.
The knitted nets made of mono-filament yarns in this research had high electro-static cling results. This indicates that they could easily build up a charge and attract airborne charged particulates to their vicinity. These particulates could be damaging to the fragile object through third-party abrasion or chemical degradation.
Based on this research, nylon net has the lowest cover of all the fabrics tested, but it also is the most abrasive and had the highest warp direction stiffness. It had high stretch and was so weak that it broke during testing before growth could be assessed. The nylon fiber degrades easily in sunlight. This combination of characteristics makes it the worst choice for conservators to use as a sheer overlay for fragile and historic fabrics. The best choice of sheer fabric for an overlay is dependent on the needs of the object being conserved and the characteristics of each fabric-balancing sheerness needs with the other properties tested.
The manufacturing and retail nomenclature for the textiles in this research is not consistent with their yarn and fabric structures. For example: three fabrics in this research are marketed as tulle: two of them are raschel knits; one is a tricot knit. Two of them have hexagonal meshes; one has a diamond shaped mesh. One is made of mono-filament yarns, one of multi-filament, and one of staple-length fibers. One cannot simply write in a report that tulle was used in the treatment of an object. Fiber content, fabric structure, yam structure, and finish information also are needed to communicate the specific nature of the overlay fabric used. Even the warp or filling orientation of the overlay treatment is useful information to be included in conservation reports for replication of results or future treatments. For these same reasons, care must be taken when extrapolating the results ofthis research to other fabrics in the market-place; specific characteristics should be matched rather than just fabric names.
The finishes applied to the fabrics in this research could be a confounding factor for a number of properties-most notably stiffness and coefficient of friction.
Information about the finishes on most of the fabrics was not available. Additional research into the finishes used on sheer overlay fabrics and the affect of the finishes on fabric performance is needed. Research into the ageing properties of sheer overlay fabrics also could provide valuable information to help conservators make informed choices for overlay fabrics, particularly if the object will be exposed to light and/or changing environmental conditions. Sheer overlay fabrics are no different from other textiles in that they have complex relationships between the yam and fabric structures and their mechanical behavior. This research provides some insight into those relationships and offers conservators objective data to differentiate between fabrics. Because each fragile or historic object requiring treatment has a set of unique needs, these data are provided to assist conservators in matching the performance properties of the sheer fabrics to these needs.  bobbinet, and 6 (14%) stated that they use other fabrics.
The "other" fabrics response included one respondent each using: allusion [sic] -nylon/poly net; batiste, hand-woven linen cheese cloth; cheesecloth, melt-bond polyester non-wovens; silk to replace silk; stretch nylon; and vintage silk organza and cotton voile salvaged from nineteenth-century garments.
Silk crepeline was the fabric used most often by the conservators surveyed.
Polyester Stabiltex was the second most used, with nylon net and tulle being the third and fourth choices respectively. This is an expected result based on a review of the literature, attendance at professional conferences, and anecdotal information.
However, results may be distorted because fabrics may have several names and/or may

146
Respondents who completed the "other" category listed a variety of possible objects including archaeological textiles, ethnographic objects, and wall hangings.
Respondents also commented that they used overlays on a nwnber of accessories such as hats, fans, shoes, gloves, and stocks. Two respondents stated that they placed overlays on embroideries, and one listed tapestries and linens as well. Two respondents placed overlays on lace; one of these mentioned using sheer fabric as an underlay under lace. One respondent stated "you could use overlays on all types of textiles." The survey should have included accessories as a choice for type of object.
Respondents may have included these objects in the costume/apparel category or not have thought of them at all. Clearly, conservators use overlays on a variety of textile projects. An additional follow up question could have determined the purpose of sheer overlays for each category of objects, but the question was not asked in the survey.
Answers to the next four questions determined if conservators pre-treat the overlay fabrics to enhance the effectiveness or aesthetic quality of the final treatment.
Question four asked about the use of adhesives with overlays. Ten respondents (23.3%) had used adhesives with sheer overlays; 26 (60.5%) did not use adhesives, and 7 (16.35%) did not answer the question. Eleven conservators answered a follow up question about the types or brands of adhesives used. One person stated that it "depends on the project." One respondent reported using adhesives "only for sticky threads." Of those mentioning a specific product, five mentioned Question seven asked whether the respondents had been doing conservation work ten years ago. Over half(25 respondents, 58.1%) answered positively, while 11 (25.6%) had not been doing conservation work ten years ago and 7 (16.3 %) did not answer the question. See Figure A4. Question 7 A asked those respondents who answered yes to question 7, which type of sheer overlay fabrics they used ten years ago. The ranking of fabrics was the same as fabrics in current use, with silk crepeline being used the most frequently and Stabiltex a close second. All of the fabrics had an increase in usage with the average increase being 11.9% with a range of 7% increase for bobbinet to 18.6% increase for silk crepeline.
Respondents had an open-ended choice of "other" for the overlays used 10 years ago; answers included cheesecloth and vintage silk and cotton organza/voile salvaged from 19th-century garments. European and other sources is difficult and, therefore, the rest of the analysis will only involve responses from North America. Respondents could select one or more of the above choices and also "other," therefore the total number of responses adds up to more than the total number of respondents. Of the eleven respondents answering "other" to question nine, several had a combination of educational backgrounds. They frequently listed internships, mentorships, and apprenticeshlps. Workshops, classes, and seminars also were mentioned. Question ten asked about the geographlc location of the conservation or restoration practice. See Table A3. Most of the respondents were from the United States due the use of address lists from American member organizations as the basis for the mailing list. The higher density of museums in the northeastern U.S. may explain the larger number of conservators located there. Unfortunately, some of the U.S. respondents did not choose one of the categories given for geographic location and chose "other" instead. This means that their answers are not included elsewhere in the analysis when geography is related to fabric use. They did explain where they were from, and this information should be taken into account for the geographlc divisions offered in future surveys. The respondents were not restricted to only one response, and the nwnber of responses to this question is greater than the total nwnber of respondents. Eighteen respondents provided comments. One defined an overlay as a treatment that "is not weight bearing" and stated that fiber content is not important because "hanging behavior" is not an issue without bearing weight. Two commented that they had used overlays very few times or none at all because their practice did not have a need for it.
A private practitioner mentioned that the varied needs of the clients, and the intended end use often dictated his/her choice of overlay materials. Another stated that he/she usually chose silk over Stabiltex because the quilts he/she worked on would be moved, folded, and used frequently. A bad experience with the use of overlays on silk quilts was related in one comment: "using the overlays on quilts ... created little sacks which would hold additional bits of silks, etc, which were shattering. After a few years, this was worse looking than nothing." Respondents used the space to ask questions: "I need a source for bobbinet" and "I need to know about the stitches people use in attaching overlays, how to handle edges, etc." One respondent provided suggestions for attachment: "use very fine needles, working by hand, with magnifying glass if necessary: rarely do I turn under edges, never with netting." One respondent liked that the fact that fabrics could be stabilized with overlays "without too much sewing and over-handling." Another commented that one needed to be sure that the original textile or garment was "strong enough to withstand sewing or adhesive attachment." One mentioned that nylon and bridal net are "not inappropriate when used in a museum setting with controlled exhibition scheduling and light levels." He or she also liked the fact that overlays were available in an "enormous quantity of colors, are cost effective, and are easy to apply and take off." A respondent expressed concern that an overlay alone does "nothing to actually stabilize a fragile textile" but suggested that using an adhesive treatment with the overlay would provide the needed stability. One person complained that the cost of fabrics was so high that small museums could not afford to use sheer overlays.

Cross-Tabulations
Cross-tabulation analysis examined trends in the data. This is a useful method for describing the interactions between the various questions. Questions such as "how many conservators who practice in the northeastern United States use Stabiltex?" can be answered using cross-tabulations. No statistical analysis was performed on these data.
The respondents' criteria for choosing fabrics was crossed by each fabric used and by the category of objects conserved. For example, a cell in a table was created by counting the number of conservators who use silk crepeline and also chose sheerness as a criteria. Because conservators could choose more than one answer to every question, none of the rows or columns adds to the total number of respondents.
Cross-tabulations also were used on the survey responses to understand changes in fabric choice over time, geographic influence over fabric choice, educational influence over fabric choice, and workplace influence over fabric choice.
The cross-tabulation analysis showed that conservators who work on all categories of objects except upholstery choose silk crepeline most frequently. See· Table A4. Stabiltex is the most frequent second choice. Upholstery conservators use Stabiltex as their first choice, with silk crepeline as second choice. Nylon net is the third choice for all objects but is used less than half as often as the first two choices.
Costume conservators use georgette more often than other object conservators do.
Quilt and costume conservators also use  Eleven criteria for choosing fabric were given as choices in the survey plus "other." Respondents could select as many criteria as they felt were important. See Table A5. Fiber content, sheerness, color, and hand were the criteria identified as The same cross tabulations can be evaluated from a different perspective, Those respondents who said that sheerness was an important criterion chose silk crepeline and Stabiltex most often with nylon net and tulle forming the next tier of choice. Silk crepeline and Stabiltex were chosen when fiber content was a criterion.
The fiber content results in this survey may be somewhat distorted as the bobbinet, tulle, and Stabiltex choices on the survey did not specify a fiber content while the silk crepeline, polyester georgette, and nylon net did specify a fiber. Silk crepeline and Stabiltex were used most by those who stated that color was an important criteria, with tulle and nylon net again being a second choice. Stabiltex is available in a nine colors and is difficult to dye. Nylon net is available in numerous colors and is dyeable.
Silk crepeline is available in only three colors but is dyeable. Tulle is available in several fiber contents, and its dyeability will be dependent on its fiber content.
Cross tabulations also compared the criteria used in making choices to the type of textile being conserved. See Cross-tabulations compared current fabric choices with fabric choices of conservators practicing ten years ago. See Table A7. The survey format did not allow analysis of how individual conservators fabric usage habits have changed over time, but the cross tabulations indicate trends in overall fabric usage. Conservators, who used silk crepeline ten years ago, continue to favor silk crepeline overwhelmingly today.
They also use Stabilitex frequently. Those who used Stabiltex ten years ago continue to use Stabiltex and also silk crepeline. Nylon net users ten years ago continue to use nylon net, but now choose silk crepeline more than nylon net and also use Stabiltex and tulle. Conservators who used tulle ten years ago choose slightly more silk crepeline than tulle now; they also choose Stabiltex and nylon net. Those who worked with georgette ten years ago choose more silk crepeline and choose nylon net and tulle equally with georgette. Those who used bobbinet ten years ago, use slightly more silk crepeline, Stabiltex, and nylon net today and use tulle and georgette equally with the bobbinet.
The demographic data were compared to fabric choice using cross-tabulations.
Fabric choice was compared to workplace, educational background, and the geographic location of each conservator's workplace.
When comparing fabric choice to workplace, conservators in private practice When fabric choices were compared to educational background, silk crepeline was again the most used fabric. See Figure A9. Those conservators who trained at Universities in the United States or Canada chose silk crepeline and Stabiltex most often, but indicated that they used all six fabrics in their current practice. Self-taught conservators in the United States and Canada also chose silk crepeline most often, but tulle was their second most commonly used fabric, with nylon net coming in third.
Those with a self-taught background did not use bobbinet. Those with other

Conclusion
The purpose of the survey was to determine which sheer fabrics conservators use today and if they use adhesive, paint, or dyes to enhance their effectiveness. Your participation in this web-based survey will help increase knowledge regarding physical characteristics of conservation textiles and their use in the care of historic and fragile textile objects. Answers to the survey will be used to select the specific textiles to be used in Masters Thesis research at the University of Rhode Island. Various standardized performance tests will be conducted on the textiles selected by the survey and data will be analyzed and discussed When you respond to the survey, your answers will go directly into a database of aggregate data. Your responses will remain anonymous, and the researchers will not be able to separate out any individual's responses to the survey. The analysis will be based on group data, and will not identify you or any individual's answers to the survey. Please feel free to contact me if you have any questions or concerns: donnalavallee@uri.edu Thank you very much for taking the time to answer these questions.  (Tortora and Merkel 1996).
English net A net with hexagonal meshes (Tortora and Merkel 1996).
Flannel A light or medium weight fabric of plain or twill weave with a slightly napped surface (Tortora and Merkel 1996).
Georgette A lightweight, sheer plain weave fabric made of silk or manufactured fiber (Tortora and Merkel 1996).
Knotted net Large mesh net knotted by hand or on a bobbinet machine (Tortora and Merkel 1996).
Illusion A fine sheer knitted net fabric (Tortora and Merkel 1996).
Laminate A layered fabric structure wherein one or more fabrics are bonded to a continuous sheet of material, such as polyurethane foam, by heat or adhesive (Tortora and Merkel 1996).
Maline A fine, diamond shaped, open mesh knitted net made of silk, cotton or manufactured fibers (Tortora and Merkel 1996).
Muslin A firm, plain weave cotton or cotton blend fabric available in a wide-range of qualities and weights (Tortora and Merkel 1996).
Mesh A fabric characterized by open spaces between the yarns. It may be woven, knit, crocheted, knotted or lace (Tortora and Merkel 1996).
Net A general term for an open fabric formed by weaving, knitting, knotting, crocheting or twisting yarn, thread or rope together to form a meshwork (Tortora and Merkel 1996).
Nylon net A sheer knitted net made of nylon (Picken 1985).
Print cloth A plain weave cotton, rayon or blended fabric in medium weights (Tortora and Merkel 1996).
Raschel A warp knit fabric made on a Raschel machine (Tortora and Merkel 1996).
Sand-fly net A very fine mesh warp knitted net (Tortora and Merkel 1996).
Sheer A transparent or lightweight fabric such as chiffon, crepe, georgette or voile of various constructions and yams. May be spun or filament yam, often silk or manufactured fibers (Tortora and Merkel 1996).
Stabiltex Trade name for a sheer, plain weave polyester fabric frequently used for reinforcing and backing fragile textiles (Salik, Salik, and Salik undated).
Terelene Trade name for Stabiltex, more frequently used in Europe than the United States (Salik, Salik, and Salik undated).

170
Tetex Trade name for Stabiltex, more frequently used in Europe than the United States (Salik, Salik, and Salik undated).
Tricot A warp knit fabric structure. A variation of tricot fabric is an open lace-like structure (Tortora and Merkel 1996).
Tulle A warp knit net with a hexagonal mesh made of silk, cotton, or manufactured fiber (Tortora and Merkel 1996).
Velveteen A cotton or cotton-blend fabric with a short, close filling pile cut to resemble velvet (Tortora and Merkel 1996).