Skip to navigation – Site map

HomeEGG Special IssuesEGG 3DossierAwareness in rolling, research ab...


Awareness in rolling, research about its effects and use of polyurethane memory foam

Studies about an oversize contemporary canvas painting: Notte barbara by Pinot Gallizio
Paolo Gili


Even though the rolling of paintings on canvas is a centuries-old tradition, it has not yet found scientific references for the identification of the best curve radius for the cylinder and for the most suitable packaging materials. Moreover, physical hazards connected with this risky procedure are often underestimated. The attempt to define guidelines for rolling up paintings led to the creation of a “flexometric table". Furthermore, an initial attempt to experiment with the use of two types of polyurethane foams led to the establishment of several tests which demonstrated its excellent characteristics during the rolling phase of the artworks, especially for textured ones.

Top of page

Full text

First of all, I wish to thank Antonio Iaccarino Idelson for his precious contribution towards the definition of this article, but most importantly for having always been present with patient kindness and a will for discussion. I want to express my gratitude for GAM, Galleria d'Arte Moderna in Tourin, in particular Virginia Bertone. A thank you goes to Antonio Rava for his presence and constant support. My heartfelt gratitude goes to Giorgio Scapitta, whose mathematical academic prowess restored order to my untidy calculations, and to Alexander Parise to whose professionalism I owe the timely, clear and precise important contribution to the translation of this text.


1Paintings on canvas may be distinct in two categories: paintings on stretchers and canvases without constraint. This distinction allows us to take further consideration of the materiality of the paint layers, when dealing with handling and storage. Size and weight are among the aspects that determine most of the handling difficulties. For unstretched paintings, the thickness of the paint layers becomes an issue of fundamental importance, since it determines the capacity for the work to flex, be rolled up or, on the contrary, the need of keeping it flat for handling.

F. 1 Notte barbara

F. 1 Notte barbara

Pinot Gallizio, Notte barbara, 1962.GAM, Galleria d'Arte Moderna Torino

Credits: © Paolo Gili

2Notte Barbara by Pinot Gallizio is an oversize canvas painting without a stretching device. Measuring 220 cm by 985 cm, it is part of the cycle of the Notti di Cristallo and was created in 1962. The canvas is a light cotton duck with a thin preparatory layer.  Paint and preparation layers, made of mixed media - PVC, PVAc, oil – have very variable thicknesses and mechanical characteristics. Heavily textured areas reach a thickness of 25-30 mm, while often paint is thin and brittle. The painting was in rather good condition, with some delamination and losses, and superficial grime. Still, it represents a conservation challenge because of the unevenness of the paint distribution combined with the absence of a stretcher or other supporting structure. The large size also makes it very complex for handling and for the identification of suitable exhibition and storage space. Rolling seemed to be unavoidable. Research concerning the issues related to the rolling of paintings and to the choice of the most appropriate protective materials with which to perform this delicate operation was inspired by the encounter with this work.  Besides the realization of the “flexometric table”, experiments with foamed polyurethanes focusing on the phases of rolling of textured paintings were performed.

Awareness in rolling

3The traditional method for rolling a painted canvas requires to use a rigid cylinder in order to grant solidity and support to the free canvas, which is clearly extremely vulnerable under such conditions. Establishing the correct diameter of the cylinder is generally an empirical decision. Its size must be small enough as to allow for ease in handling and large enough as to not damage the work. Furthermore the painted surface must face the outside. This practice allows for easier handling of large works and is also less expensive; however, these are the only available indications concerning the rolling of works of art. Often the free interpretation of these general guidelines has, in fact, led to serious errors such as the use of undersized cylinders and also the use of unsuitable materials for the protection of the paint layer included during the rolling process.

4Regarding the related bibliographic literature, it has proven impossible to locate any scientific reference detailing which correct curve radius should be applied to a painting that is to be rolled and there are no indications of any kind if not the ones based on empirical information. Older canvas paintings had very low thicknesses of color and for this reason their rolling could also be carried out on small diameter cylinders without causing visible damage to the paint layers. From the second half of XIXth century the painting technique has seen the use of color in much greater thicknesses but the radii of the cylinders used for these more textured works weren’t always suited to this new condition and cylinders with insufficient diameters continued to be used. Painted surface have sometimes been heavily damaged and, in the XX th  century, the same has happened to works which used new industrial materials whose characteristics are very different from traditional colors. The literature states that the diameter of the cylinders used for rolling ranges between 40 and 100 cm. In certain cases the diameter is even smaller (only 15-20 cm) if present at all. Rarely does the diameter exceed the measure of one meter.

5Not only does the diameter of the cylinder become of fundamental importance when rolling a canvas painting; it is also important to adequately protect the paint surface when the material shows greater thicknesses and morphological unevenness. The protection of the painting, in almost all cases reported in literature, requires the use of technologically advanced products - breathable and highly compatible with the original materials. However, they may not always be able to adequately perform their duty of protecting morphologically complex and/or fragile paint surfaces, since they may not be able to correctly conform to the shapes.

6As will be observed later, the ideal cylinder does not exist. This means that even if a minimum flex of a painting is never desirable, it is, however, sometimes required. Knowing the effects of rolling and the forces that are generated within the layers of the paintings in this condition is very important in order to understand certain mechanisms that are often overlooked.

The materials of the paintings and the effects of rolling on the pictorial film

7If a layer of material of any thickness is rolled, this will produce, due to the bending moment, forces that are of opposite sign but of equal magnitude on the two surfaces that delimit the layer: one side will be compressed and the other will be stretched, so that the first will be shortened and the second will extend.

Fig. 2 Bending moment

Fig. 2 Bending moment

Credits: G. Carbonaro

8At the core of the material layer there will be an ideal plane behaving as if it were not subjected to the applied force and where the value of the forces is 0. Between this level and the two respective surfaces that delimit the layer there is a gradient with increasing values of force.

9Because of the great variety of materials employed, for the most part composite, and because of the great inconsistency in their aging, which determines extreme differences in physical and mechanical behavior, within a work of art it is possible to be faced with a wide range of physical behavior of materials. They can be very ductile, moderately ductile and brittle.

10Every material, when elastically stressed, has a (small) change in shape, volume, or both. The energy required to produce this change is stored in the body as elastic energy. However, a material has a limited and defined ability to absorb energy of distortion whose trend is to change the shape but not the volume. Every attempt to increase the distorting energy applied to a body beyond  said threshold will yield firstly an elastic stretching which then reaches a point of enervation, followed by plastic deformation and, finally, by breakage. (Schodek 2005)

11Therefore, pictorial materials can have three different types of behavior when stressed. In the first case, the material is elastic enough and the applied force is not able to cause permanent deformation, so when the applied force is removed there is a complete recovery of the original condition.

12In the second case, the force exceeds the yield point and causes enervation. When the applied force is removed the pictorial film maintains a partial or complete deformation. Additionally, a partial detachment of the stressed layer or of portions of material close to this area may occur.

13In the third case the applied force is such that the pictorial layer exceeds both its temporary and permanent abilities for elastic deformation, and so it breaks. The occurring effects are, for example, crimping, where every crack is a stress release generated by forces affecting the material. Such cracks usually occur at a right angle to the direction of the force and will continue to function as stress release points also during future force applications.

14These tensions may occur because of internal material issues (shrinking during the drying phase, thermal and humidity variations) or external causes (collisions or bending).

15The materials comprising a canvas painting, the differences in composition, method of preparation / application, different aging conditions even in the same layer (areas exposed to UV rays and to the atmosphere, while others remain  more protected) lead to very complex situations where any difference between the layers, even if consistent with each other, acts as a preferential level of detachment or sliding, which can occur in the case in which delamination or a partial weakening are already present, thus causing more serious damage in the immediate surrounding areas.

16The value of the compressing and stretching forces show an exponential progression graphically represented by a parabolic curve.

The "Flexometric table"

17The rolling operation can be physically described by understanding three data: the thickness of the art work, the diameter of the cylinder and the consequent stretching of the painted surface. By placing these pieces of data in relation with each other on a Cartesian diagram, it is possible to graphically describe the type of stretching that a painted surface is subjected to according to its thickness and to the diameter of the cylinder around which it is rolled.

18In order to establish the value of the stretching it is necessary to calculate the circumference of the cylinder taken into consideration for rolling and to subtract it to the circumference that the painted surface, of a certain thickness, creates around the same cylinder. The difference between the two measurements, divided by the cylinder’s circumference yields the value by which the painted surface will be lengthened during its first revolution around the cylinder itself. The stretching value for subsequent revolutions can be calculated by repeating the same operation and by considering the cylinder’s diameter as increased by the value of thickness of the artwork which has already been rolled around it. This stretching value will become smaller and smaller with each subsequent revolution. For this reason it is important that the best compromise for rolling be assessed with reference to the first revolution of the artwork around the cylinder, since all following revolutions will be subjected to decreasing stresses.

19With the help of mathematician Dr. Giorgio Scapitta, it was possible to write the general equation describing the relation between these three pieces of data regardless of any numerical values. The objective was to obtain, in the most comprehensive way possible, an equation able to describe these relations and which could be applied to any real example.

20For the sake of completeness in understanding the logical steps taken, the following are principles used in order to structure the equation, for the mathematical definitions please refer to the thesis.

21The diameter of the cylinder, the thickness of the artwork and the circumference of the cylinder, reduced to their simplest descriptions, have all been taken into account.

22A geometric evaluation was then performed, taking into consideration the condition of an artwork which had already been rolled over a cylinder whose circumference was equal to the length of the artwork itself, in order to describe a single revolution around the cylinder.

23Next, the “diameter of canvas” and “diameter of paint layer” values were defined as the measure of the diameter of the cylinder increased by the added thickness, rolled over itself, of the artwork and calculated twice.

Fig. 3 Section of the cylinder with first revolution of the artwork and measures taken into consideration

Fig. 3 Section of the cylinder with first revolution of the artwork and measures taken into consideration

Credits: © Paolo Gili

24Having obtained the measurements of the circumference of the cylinder and of the artwork or of its parts, their difference was then calculated.

25The calculated variance, divided by the measure of the circumference of the cylinder describes the stretching to which the paint layer of the artwork was subjected during that single revolution.

26It is possible to also take into consideration the variance occurring between other intermediate layers of the artwork.

27All calculable values of stretching can be summarized with a single equation which describes them all.

28Thus, it became possible to obtain the stretching coefficient of the painted layer with respect to the circumference of the cylinder.

29αPC = 2(a+b)/x

  • αPC” indicates the stretching of the paint layer.

  • The numerator “2(a+b)” allows to calculate the circumference of the outermost layer of the painting upon its first revolution around the cylinder. The parentheses can include the sum of all the measures of the various layers of the painting or only a partial piece of information.

  • The denominator "x" indicates the circumference of the cylinder.

30By drawing on the psychrometric diagram, which graphically represents atmosphere properties under various conditions, it was possible to define a diagram able to show at the same time the principal data addressing artwork rolling and which are useful in order to correctly understand the necessary practical choices.

  • The x-axis corresponds to the values (expressed in mm) relative to the diameter of the cylinder.

  • The y- axis corresponds to the stretching values to which the paint layer is subjected during its first revolution around the cylinder. These values are expressed in mm for each linear mm of paint layer and in a stretching percentage.

  • Each of the graph’s curves represents a different thickness.

Fig. 4 The flexometric table

Fig. 4 The flexometric table

Table for the identification of the best compromise between the diameter of the cylinder and the ability to handle it in relation to its size

Credits: © Paolo Gili

31The flexometric table is the result of this work and is intended be a useful tool for the identification of the best compromise between the diameter of the cylinder and the ability to handle it in relation to its size. It also allows to quantify and to understand the effects that this operation has on artworks.

32The stretching value yielded by the table must be placed in relation with the material comprising the painting. The same stretching value percentage will have different effects on a recent or an ancient painting, perhaps made with the same materials but in different moments. Real quantification of stresses and, therefore, of damage, must allow for a huge amount of variables which simply cannot be all managed and evaluated properly. There are specific studies addressing the failure mechanisms of painting materials and their physical characteristics, for example the extensive research performed over the years by Marion F. Mecklenburg (Mecklenburg 2007). These results may better contextualize the employment of the flexometric table.

Protection and support of the paint layer during rolling

33A canvas painting rolled onto a cylinder is mostly handled and stored horizontally. This condition subjects the artwork to physical stresses, each of which is very different from the other and which can vary significantly according to the type of rolling performed and to the materials employed to isolate each revolution of the canvas around itself.

34The main issues when rolling up paintings occur when the paint layer shows considerable thickness and texture. If brush strokes of color are in relief, which are very thin and fragile, they can be easily damaged under the slightest pressure or stress.

35While great care should always be granted to the choice of the most appropriate size of cylinder, another aspect of equal importance must also be considered, regarding the overlapping of the work on itself at each complete revolution around the cylinder. It is necessary to identify the most suitable material to protect and support the painting’s surface.

36Rolling is like stacking multiple separate paintings. Each layer presses on the one below. The top of the brush strokes support the weight of the other layers. After a certain load or pressure the most fragile and higher brush strokes break or deform and flatten permanently.

  • 1  These are used in particular to protect thicker and more fragile brush strokes, but they are too r (...)

37The two surfaces of the painting must also be kept ventilated. In most cases the pictorial layer is protected with layers of acid free paper, nonwoven fabric, bubble wrap, Melinex, lately with Tyvek or more rarely with Volara and other types of polyurethane foam1. The latter are usually employed in order to protect the thickest and more fragile brush strokes, but have shown to be too rigid and undersized – in thickness – to properly adapt to the paint layer. These devices isolate the paint film from contact with the next layer, but do not always allow for correct protection and ventilation of the canvas’ surfaces. The use of these non-deformable materials (films which are relatively rigid) implies a further problem in case the painting’s surface shows substantial irregularities in its texture.

Fig. 5 Section of a sample

Fig. 5 Section of a sample

Section of a sample showing two rolled layers of a thickly textured and incorrectly protected painting

Credits: © Paolo Gili

38Illustration 5 shows what happens to irregular paint layered with high brush strokes during rolling. They clearly show two problems: the weight of each single revolution which the artwork performs around itself presses unevenly over underlying layers weighing down mainly on higher brush strokes. Furthermore, after the first complete revolution around the cylinder, the canvas no longer has an even support. The use of insulating rigid or semi-rigid materials such as Tyvek or unwoven fabric can only worsen the issue of uneven weight distribution, causing more stress and possibly more damage to the paint surface.

39Various stresses occur. The paint layer is subjected to uneven compression since the artwork is supported mainly by the higher and thicker brush strokes, overloading them and therefore, incurring the risk that these parts may deform or rupture. These points of support are the cause of unfavorable pressure not only against the layer of color, but also against the entire structure of the artwork. The lack of a continuous and flat support, caused by the presence of a morphologically complex surface, determines a curvature of the canvas which uncontrollably extends in the parts which are convex and dangerously compresses in the parts which are concave.

Proposal for a new material intended for the rolling of heavily textured paintings: memory foam layer

40In order to evenly distribute the weight of the painting layers on the ones below and to protect the strokes of color, it is necessary to fill the gap between the thinner and less prominent layers and the higher ones so as to obtain a flat and continuous surface over which to lay the next layer of painting.
What kind of technical and physical characteristics are required for a material so that it can be inserted between each revolution which the artwork performs around itself?

  • Perfect morphological adaptation to the pictorial surface

  • Compatibility with the materials constituting the artwork

  • Able to allow adequate ventilation of painting surfaces

  • Support for the thinner and more fragile pictorial material

  • Even distribution of weight over the entire paint film of the lower layers

  • It must create a sustaining surface which is, even, supported and stiff enough to remain planar

  • Lightness, adaptability to special measures.

41In order to achieve these results, it was decided to employ a special category of materials, specifically polyurethane foams. They are a very broad family of polymeric materials characterized by the presence of urethane linkages within their molecular chains. The chemical reaction required for their synthesis was discovered in 1947 by Bayer and their employment has seen a substantial acceleration with various applications in the most diverse fields. These materials have already been employed in restoration but seldom for the operations of rolling-up of paintings. In this specific field they are used in very thin layers (a few millimeters) and without making focused choices on specific densities intended for the paintings to be protected.

42There is an almost unlimited variety of polyurethane foams which differ in many respects: density, the ability to be breathable (i.e. to have open or closed pores), the size of the pores and other characteristics.
Research has also considered memory foams; viscoelastic open cell polyurethane foam which possesses excellent characteristics for adapting to different types of compression, high capacity for absorbing vibrations, continuous support and they are able to significantly attenuate the points of pressure. The open pore structure featured also by this category of material allows for the passage of air within itself, so as to avoid air and humidity stagnation in direct contact with the materials comprising the artwork.

Fig. 6 The employment of a continuous layer of polyurethane inserted between the revolutions of a rolled heavy-texture painting

Fig. 6 The employment of a continuous layer of polyurethane inserted between the revolutions of a rolled heavy-texture painting

Credits: © Paolo Gili

43The proposal is to use a layer of polyurethane foam of suitable thickness, technological characteristics and equal dimensions to those of the artwork only for the rolling and transport phases.
Before packing the polyurethane layer can be laid directly on the surface of the painting so that when the painting will be completely rolled on its cylinder the foam will be between each layer that the painting generates rolling up around itself. Besides acting as an insulation, its main task lies in the double behavior with regard to the two layers of the artwork which it separates: one surface of the polyurethane layer is able to morphologically adapt to the pictorial film filling the empty spaces and supporting every single color stroke while the opposite side remains planar and it is itself a support for the next revolution of the painting.

Fig. 7 Simplified stratigraphy when using a panel of polyurethane

Fig. 7 Simplified stratigraphy when using a panel of polyurethane

Credits: © Paolo Gili

44By correctly assessing the characteristics of the paint layer, by measuring its thicknesses and knowing the length of the painting and the size of the cylinder, it will be possible to determine the number of revolutions that the artwork will have to perform in order for it to be completely rolled and thus to establish the overall size of the panel of polyurethane. The even thickness will be suitable for maximum stresses and will be redundant towards the outside, since it will have to support less weight with respect to the centre of the roll.

Experimental phase

45The first phase of this modest research lies in the approach to an evaluation concerning possibilities and  methods for using polyurethane foam. Besides supplying certain initial useable data, the study herein carried out has certainly allowed to understand the complexity of a topic which has gained little attention in an attempt to articulate and define a future experimental procedure able to not ignore any aspect worthy of being further studied.

46Even if the number and variety of polyurethanes acquired and tested is not enough and remains lacking in terms of sample uniformity of measurements and uniformity of technological characteristics, it was nonetheless decided to proceed with the experimental phase in order for new data, able to generate further discussions, to emerge.

Polyurethane foams tested

47The polyurethane foams tested are made in Germany by Metzeler Schaum Inc. Polyoles and polyisocyanates are generated in crude oil production, it was impossible to now have other more specific information on the chemical formulation with which the polyurethanes are produced. The company website describes that PUR soft foams are environmentally compatible products. They are also perfectly harmless physiologically and contain no toxic heavy metals. They do not contain any other materials considered harmful today, such as nitrosamine, formaldehyde, asbestos, PCB (polychlorinated biphenyls), PCP (pentachlorophenol), styrole or vinyl chloride. (

48In order to verify the employment possibilities of this family of materials, through the identification of the one showing the best performances, we prepared several samples so as to assess the effects that certain pressures have on different types of paint surfaces, by interposing various thicknesses and densities of polyurethane.

49The paint surfaces were prepared on a very thin simple weave cotton canvas, replicating the type of support used by “Notte barbara”, but which was especially light and very flexible. Three different kinds of samples were prepared according to the kind of test on the same kind of support.

  • 2  The size of the imported polyurethane foam panels are 60 meters in length with variable sections o (...)

50The experiment considered 11 types of foams: 2 foams with viscoelastic characteristics and 9 standard polyurethane foams. These materials were provided by the Dorsal company in Corbanese in the province of Treviso. The Company buys their polyurethanes in Germany in blocks of several cubic meters2 which are then cut and processed according to the requirements of the specific products.

51Each type of polyurethane foam is characterized by 4 digits of numeric data. The first two refer to the density in Kg/m3, the final digits must be divided by 10 to get the value for the compressive strength expressed in kilopascals (KPa). For example a polyurethane such as - memory 60.28 - has a density of 60 Kg/m3 and 2.8 KPa compressive strength which is approximately equivalent to 285,5 Kg/m2.

Fig. 8 Polyurethane foams tested

Fig. 8 Polyurethane foams tested

Credits: © Paolo Gili

52All foams were supplied pre-cut in the shape of a parallelepiped of different thicknesses, but each with a surface area of 100 cm2 which was placed in contact with the sample.

53The tests was performed in three series and was carried out on a flat and rigid surface upon which the canvas sample was placed and, in turn, on which polyurethane sample was placed over which, lastly a rigid board was located in order to evenly distribute the weight. The foams were subjected to four known weights: 0.5 Kg (5 gr/cm2  = 0,5 KPa); 1 Kg (10 gr/cm2 = 1 KPa); 1,5 Kg (15 gr/cm2 = 1,5 KPa); 2 Kg (20 gr/cm2 = 2 KPa).

First set of tests : surface protection

54The first set of tests was performed in order to verify which foam was able to guarantee the best protection of the more fragile and thin layered paint, achieving a pressure of up to 2 KPa/cm2.
The first series of textures were realized with unbound plaster so as to obtain a very fragile layer. Two types of coating were produced: the first was realized with a spatula cut on one side with the desired shape so as to leave continuous 5 to 20 mm high crests of material on the support during the pulling of the soft plaster. The second coat was applied to the canvas as single plaster spikes, similar to small cones, 5 to 20 mm high.

55The results showed that all the samples of the tested polyurethane have excellent ability for weight distribution and protection of the thin plaster crests of the sample. The thinnest tested polyurethane had a thickness of 25 mm and even in these cases the results were very satisfactory.

Second set of tests : morphological adaptability

Fig. 9 Second set of tests

Fig. 9 Second set of tests

Morphological adaptability

Credits: © Paolo Gili

56The second set of tests was designed to test and assess how the various foams would adapt to a morphologically complex thick texture coating. The set of approved samples for this test was made of colored MAN series PVC manufactured by the Plastocoat company in Torino between 1958 and 1962 and used by Gallizio for the Notte Barbara canvas. Coating was performed in an attempt to replicate the same complex morphology and texture as seen on the painting.

57The calculations of the foams to adapt were carried out by photographing the sections of the samples under load and line-measuring the surface subjected to the load and the contact surface which the polyurethane was able to effectively cover.

58The morphological adaptability test showed significant differences among the tested polyurethanes, in particular for the category of the standard polyurethane foams and visco-elastic polyurethane foams. These latter showed excellent adaptability, ranging between 62% and 96% even at a pressure of just 1 KPa/cm2.

Fig. 10 Results for polyurethane adaptability to the sample surface

Fig. 10 Results for polyurethane adaptability to the sample surface

Credits: © Paolo Gili

59According to the collected data (as shown in ill. 10) we chose the polyurethane which demonstrated the best adaptability performance.

60Memory 50.15, with a thickness of 10 cm, possesses a 96% adaptability at a pressure of  1 KPa/cm2  which revealed itself to be the highest of all tested foams and achieving 97% at 2 KPa/cm2. Among the non-viscoelastic foams with an adaptability of 89% at 1 KPa/cm2, the best was 23.09, which has the same very low density characteristics (only 2.3 Kg/m3 ). The performance of the foam itself does not improve dramatically when subjected to double the pressure, achieving only  89.65%. Unfortunately, the limited amount and size variety of the tested polyurethanes did not allow to fully understand which physical characteristics of these materials determines the best adaptability to a morphologically complex surface.

Third set of tests : support

61The third set of tests was intended to verify the response of different polyurethane foams sandwiched between two canvases: this is, in fact, the main purpose for which these materials were taken into consideration in this research.

62The third material employed for the making of the samples is an acrylic gel (Liquitex® matte gel medium, acrylic polymer emulsion. Made in France by Liquitex Artist Material) used in painting to give texture to the strokes of color. The choice for this material stems from the fact that, once it is dry, it is compact and rigid such as a coat of oil paint which requires longer periods for polymerization.

63The application of the gel on the canvas followed the same procedure as the plaster samples, by using a spatula with a pre-cut side. In this case the morphology took into account an alternation of texture thickness allowing for different conditions: high crests very close or very far apart, lower crests and very slight undercuts

Fig. 11 Tests

Fig. 11 Tests

Tests of various polyurethanes sandwiched between the layers of the artwork
Credits: © Paolo Gili

64Illustration no. 9 shows two kind of polyurethane foam in compression between two revolutions of a painting rolled up on itself. The first one used a low-density and low resistance polyurethane foam seems to have a poor support for the second level at a pressure of 1 KPa/cm2 and a satisfactory morphological adaptability to the surface of the first level. If the pressure is increased to 2 KPa/cm2 the support is very poor with respect of the second level, even it maintains a good morphological adaptation with respect of the first layer. The second one used a well-supporting foam and the result obtained at a pressure of 2 KPa/cm2 is effectively satisfactory, but to the detriment of a better conformation of the polyurethane to the first layer of the artwork.

65The results of the two tests allow us to understand that you cannot get the best result of morphological adaptation and the best support for a textured layer of paint using only one type of polyurethane foam. It becomes therefore necessary to create a double layer able to account for both needs during the rolling process.

66However, we must clarify that a single layer of foam, in the case of a painting without any criticalities in its raised surface, may have excellent conformation characteristics and may also be able to adequately support the various layers of the painting.

Fourth set of tests : adaptability and support

67The aim of  this phase of the research was to identify the best combination of a first layer of polyurethane in direct contact with the paint layer –(the 50.15 memory foam was chosen in the test’s second phase) – and a second foam intended to provide support for the artwork’s subsequent revolutions.

68In order to perform this test it was chosen to assess the results of all available high density polyurethanes (the series featuring numbers 60, 50 and 40; respectively 6 Kg/m2, 5 Kg/m2, 4 Kg/m2) at pressures of 2 KPa/cm2 and 3 KPa/cm2.

69The test was performed on a flat and rigid surface, with the canvas sample prepared with the gel laid on top of it. The 50.15 memory polyurethane sample was in turn placed on top of it, followed by the three series of standard high-density polyurethanes, each one covered by a second canvas sample prepared with the gel. Finally, pressure was applied to the entire stack. Load methods were differentiated: in one case the pressure was applied through a rigid surface, directly in contact with the gel coating, while in the second case by interposing polyurethane foam able to evenly distribute the applied pressure over the entire surface.

70Thus, the standard polyurethanes are between the memory layer and the sample canvas support which emulates the second layer of a painting rolled up around itself. The test verified the ability of standard polyurethanes to adequately support the second canvas sample at different pressures and under different load methods.

Fig. 12 Fourth set of tests : adaptability and support

Fig. 12 Fourth set of tests : adaptability and support

Adaptability and support

Credits: © Paolo Gili

71The best results were obtained with 50.40 at a thickness of 3 cm, which appears to have excellent properties on both surfaces, with yielding characteristics at the interface between the two polyurethanes (a sign of proper distribution of pressure and good interaction between the two types of polyurethanes able to correctly operate in conjunction) and excellent planar support with respect to the second layer of the artwork.

Further research

72Unfortunately, the limited amount and size variety of the tested polyurethanes did not allow to fully understand which physical characteristics of these materials determines the best adaptability to a morphologically complex surface.

73An initial first topic of further research should be addressed in order to understand which technological characteristics of non-viscoelastic polyurethanes and viscoelastic (memory) polyurethanes are the ones determining their optimal performances in adaptability to a textured paint layer.A second topic addresses the reason for the different behaviors observed at the interface between two foams subjected to pressure during the fourth set of tests.

74In order for the research to be even more through, it would require to increase the number of cuts, in terms of thickness, of the polyurethanes employed during the tests, so as to allow for greater uniformity in result comparison. Experimentation performed up to this point is not completely exhaustive, since it is still lacking sufficient data to evaluate what happens to paint which remains for prolonged periods of time in contact with polyurethane foam material. In fact, regardless of being a material with excellent technological qualities, also releases phthalates as it ages.


75The first part of this work attempts to analyze and understand the level of knowledge and awareness of actions addressing the technique for rolling canvas paintings.

76The objective was to find a simple and visual method to understand the effects of rolling. Was devised a flexometric table, which coordinates and collects in a single graphical space the three pieces of data which must be accounted for during rolling. The flexometric table is a modest reference guide intended to provide indications in order to define a proper rolling procedure and to be aware of the implications when making the choices relative to this operation.

77The second part of this work intended to further analyze the rolling operation of canvas paintings addressing specifically those paintings featuring morphologically complex, thick and delicate surfaces. We attempted to describe and understand the conditions of a rolled up painting and, more importantly, its weak spots. A test phase, even if incomplete and imprecise, allowed to outline some encouraging preliminary results concerning the use of polyurethane foams as rolling material. They demonstrated that they have excellent properties in protecting the paint layer, even in very complex situation, thanks to their excellent ability to adapt to irregular coatings.

Top of page


AA. VV. (1991). Conservare l’arte contemporanea, Firenze 1992

AA. VV. (2012). Extra large, Grimaldi forum Monaco, Parigi

Berger, G. A. - Russell, W. H. (2000). Conservation of paintings, Research and Innovations, London

Bordini, S. (2007). Arte contemporanea e tecniche. Materiali, procedimenti, sperimentazioni, Roma

Capriotti, G. - Iaccarino Idelson A. (2004). Tensionamento dei dipinti su tela. La ricerca del valore di tensionamento,  Provincia di Viterbo

Chiantore, O. - Rava A. (2005). Conservare l’arte contemporanea: problemi, metodi, materiali, ricerche,Milano

Ciappi, O. - Ciatti, M. (1996). La conservazione dei dipinti su tela: esperienze e innovazioni per dipinti di grandi dimensioni in “OPD Restauro”, n. 8, 159-164

Ciatti, M. - Frosinini C. (2004).Restauri e ricerche. Dipinti su tela e tavola, Firenze

Maltese, C. (1990). I supporti nelle arti pittoriche: storia, tecnica, restauro, Milano

Matteini, M. - Moles, A. (2003). La chimica nel restauro. I materiali dell’arte pittorica, Firenze,

Mecklenburg, M.F. (1991). Art in Transit: Studies in the Transport of Paintings, London, National Gallery of Art

Mecklenburg, M. F. (2007). Meccanismi di cedimento nei dipinti su tela:approccio per lo sviluppo di protocolli di consolidamento, Padova, Il Prato

Rey, S. M. - Blay, V. G. - Agusti, M. C. (2010). Congreso International de restauracòn de pinturas sobre lienzo de gran formato, (Valencìa, 26-28 October 2010), València

Schodek, D. L. (2005). Strutture, Patron

Scicolone, G. (1993). Il restauro dei dipinti contemporanei, Firenze

Woodcock, S. (2005). Big pictures. Problems and Solutions for Treating Outsize Paintings, London

• Metzeler;; March 15th, 2013

• Sonotube;; July 15 th, 2012

•Volara;; July 10 th, 2012

Top of page


1  These are used in particular to protect thicker and more fragile brush strokes, but they are too rigid and thin to adapt to the paint surface in the best way.

2  The size of the imported polyurethane foam panels are 60 meters in length with variable sections of 80 or 100 x 212 cm

Top of page

List of illustrations

Title F. 1 Notte barbara
Caption Pinot Gallizio, Notte barbara, 1962.GAM, Galleria d'Arte Moderna Torino
Credits Credits: © Paolo Gili
File image/jpeg, 1.1M
Title Fig. 2 Bending moment
Credits Credits: G. Carbonaro
File image/jpeg, 156k
Title Fig. 3 Section of the cylinder with first revolution of the artwork and measures taken into consideration
Credits Credits: © Paolo Gili
File image/jpeg, 364k
Title Fig. 4 The flexometric table
Credits Table for the identification of the best compromise between the diameter of the cylinder and the ability to handle it in relation to its size
File image/jpeg, 1.1M
Title Fig. 5 Section of a sample
Caption Section of a sample showing two rolled layers of a thickly textured and incorrectly protected painting
Credits Credits: © Paolo Gili
File image/jpeg, 1.0M
Title Fig. 6 The employment of a continuous layer of polyurethane inserted between the revolutions of a rolled heavy-texture painting
Credits Credits: © Paolo Gili
File image/jpeg, 456k
Title Fig. 7 Simplified stratigraphy when using a panel of polyurethane
Credits Credits: © Paolo Gili
File image/jpeg, 952k
Title Fig. 8 Polyurethane foams tested
Credits Credits: © Paolo Gili
File image/jpeg, 148k
Title Fig. 9 Second set of tests
Caption Morphological adaptability
Credits Credits: © Paolo Gili
File image/jpeg, 1.0M
Title Fig. 10 Results for polyurethane adaptability to the sample surface
Credits Credits: © Paolo Gili
File image/jpeg, 208k
Title Fig. 11 Tests
Credits Tests of various polyurethanes sandwiched between the layers of the artwork Credits: © Paolo Gili
File image/jpeg, 1004k
Title Fig. 12 Fourth set of tests : adaptability and support
Caption Adaptability and support
Credits Credits: © Paolo Gili
File image/jpeg, 1.0M
Top of page


Electronic reference

Paolo Gili, “Awareness in rolling, research about its effects and use of polyurethane memory foam”CeROArt [Online], EGG 3 | 2013, Online since 10 May 2013, connection on 06 May 2021. URL:; DOI:

Top of page

About the author

Paolo Gili

Following his diploma earned at the art college, he graduated from the University of Torino in Preservation and Conservation of Artistic heritage in the Centro per la Conservazione e il Restauro dei Beni Culturali “La Venaria Reale” specifically in: painted works on wood and canvas; artworks sculpted in wood; wooden furniture and structures; artworks in processed synthetic materials, assembled and/or painted. He has been working since 2004 mainly as a restorer of ancient and contemporary works of art, researching the field of multispectral photographic diagnostics. This latter interest continues today with further studies of computational photography. Contact:

Top of page


Licence Creative Commons
CeROArt – Conservation, exposition, restauration d'objets d'arts est mis à disposition selon les termes de la licence Creative Commons Attribution - Pas d'Utilisation Commerciale - Pas de Modification 4.0 International.

Top of page
  • Logo DOAJ – Directory of Open Access Journals
  • OpenEdition Journals
Search OpenEdition Search

You will be redirected to OpenEdition Search