The Gallery holds the most extensive collection of portraits in the world. Search over 215,000 works, 150,000 of which are illustrated from the 16th Century to the present day.

Advanced Collection search

‘Three-quarters, kit-cats and half-lengths’: British portrait painters and their canvas sizes, 1625-1850

1. Introduction

The chief canvas sizes for painting, wrote Thomas Page in The Art of Painting in 1720, are ‘the three quarter’d and half length (the first show the Figure to the Waste, the latter below the knees)’.[1] It never ceases to surprise those coming fresh to the study of British portraiture that for more than two centuries what we now usually describe as a half-length portrait, size 30 x 25 ins (76.2 x 63.5 cm, at 2.54 cm to the inch), was known as a ‘three-quarters’, while the three-quarter length, size 50 x 40 ins (127.0 x 101.6 cm), was called a half-length. Why were standard sizes adopted? Why these particular sizes and not others such as those used on the Continent? And what effect did these sizes have on the practice of the portrait painter’s art? It will be easier to find answers to these questions if we investigate the nature and supply of canvas (section 1) and then trace how the range of standard sizes and the corresponding terminology evolved (sections 2 and 3).

1.1 The nature of canvas

In England the great majority of oil paintings known from the 16th century were done on panel. While canvas had long been used for decorative work, it was not until the end of the century that it came into common use as a support for easel paintings, apparently coinciding with the growing taste for large whole-length portraits for which a panel would have been less practical. [2] An example is Marcus Gheeraerts’ Queen Elizabeth I, c.1592 (National Portrait Gallery), the so-called Ditchley portrait, an exceptionally large whole-length painted on a single piece of canvas, size 95 x 60 ins (241 x 152 cm). [3] Panel supports remained in common use for smaller paintings until the late 1620s and it was only in the following decade that canvas became the standard support for portrait painting in England. Panel and canvas sizes used by artists in England before the 1620s will be the subject of forthcoming studies associated with the ‘Making Art in Tudor Britain’ project at the National Portrait Gallery. [4]


List of canvas sizes. From The Artist’s Repository and Drawing Magazine, c.1785 (National Portrait Gallery Library).

Fig.1 List of canvas sizes. FromThe Artist’s Repository and Drawing Magazine, c.1785 (National Portrait Gallery Library).

Canvas is cloth made from flax (when commonly called a linen canvas) or on occasion from cotton, hemp or jute. [5]Flax and cotton canvases are usually more finely woven than those of

hemp or jute. Until the late 18th century a painting on canvas was often simply described as being on cloth: ‘Picture-cloths are of several sizes’, reads the description in The Artist’s Repository in 1785 (fig.1). [6] Canvas is of course not exclusively a cloth for painting pictures; in 1828 John Trumbull, an American artist trained in England, described his paintings for the Capitol in Washington DC as being painted on a ‘linen cloth, whose strength and texture is very similar to that used for the topgallant-sails of a ship of war’. [7]

Most linen was imported from abroad, as Daniel Defoe noted in his account of English commerce in the 1720s. [8] A little handbook of 1695, The Merchant’s Ware-house laid open, serves to show the huge range of linens and related materials which were available in a wide variety of widths in the late 17th century, whether Continental, English or Irish in origin. [9] The import of linen was subject to import duties that were increased several times between the late 17th and the late 18th centuries. [10] As a result during the course of the 18th century Irish and Scottish linens made considerable inroads into the English market, replacing foreign imports, and linen began to be produced in quantity in England itself. [11]

Canvas as a painting support came in many grades. In the late 1670s and early 1680s the portrait painter Mary Beale was using a very wide range of supports for her portraits, as we know from her husband’s notebooks: Royston sacking and fine Oznabrug occur frequently but there are also references to coarse and fine canvas (the fine canvas was used for small pictures), onion bag (again for small pictures), bed-ticking, flaxen cloth, Dutch cloth and Gentish Holland. [12] By the second quarter of the 19th century, three types of canvas were commonly stocked by artists’ colourmen in London: plain cloth, Roman and ticken, the last being of a twill weave favoured by many portrait painters. [13]

1.2 The supply of canvas

The width of cloth, whatever the type, was traditionally specified by the quarter yard (22.9 cm) in England. The yard was 36 ins (3 feet, 91.5 cm). At one period or another from mediaeval times onwards widths were fixed by legislation variously at 4, 5, 5 ½ , 6, 6 ½ and 7 quarters, although practical considerations of cloth production meant that such requirements were sometimes ignored. [14] In the 1830s Roberson & Miller were offering canvas in six yard pieces of widths 27, 30, 36, 39, 42, 45, 54, 62, 74 and 86 ins (68.5 to 218.5 cm). [15] Despite the repeal of almost all restrictive legislation in 1831, cloth widths of 36, 45 and 54 ins (4, 5 and 6 quarters) endured until the adoption of metric sizes in Britain in the 1970s. [16] It is worth adding that in the early 20th century one American importer of English canvas, Favor, Ruhl & Co of New York, noted ‘Every roll measures, for Stretching purposes, about 2 inches more in width than marked in the list’. [17]

As for cloth length, the measure adopted was commonly the yard (36 ins, 91.5 cm) although the ell (45 ins, or 5 quarters, 114.5 cm) was often used in the 16th and 17th centuries. [18] ‘A picture of A Lady an ell long’ features in one of the documents concerning Van Dyck’s estate in 1649, while ‘Two Ell pictures over ye doors of Van Dikes’ are included in a list of pictures repaired for the Earl of Clarendon in c.1683-5. [19] The term was used by Charles Beale in 1677 when buying materials for his wife, Mary, e.g. ‘20 ells & ¼ of good canvas at 20d the ell’. [20] As late as 1743 George Vertue describes a set of landscapes, apparently by Cornelis Huysmans, as ‘about an ell by ¾’. [21] A very late instance of its use can be found in the listing of an ‘Ell Hatchment’ canvas in Jabez Barnard’s catalogue as an artists’ colourman of about 1860. [22]

How in practice were canvases for artists cut from the roll of cloth and then prepared? Some of the most revealing insights are contained in Charles Beale’s remarkable diary notebooks for 1677 and 1681 of work that he undertook for his wife, Mary. [23] They contain the most detailed record in Britain of the process, from the purchase of the canvas in rolls, the way it was divided into individual canvases, the making up of the primer, the preparation of the canvas, the use of special straining frames for the actual priming process, the use of ordinary straining frames for stretching the canvas once primed, and the process thereafter of storing and then using the canvas for painting. What is of further interest is that despite having the freedom to make up canvases in whatever sizes he and his wife chose (unlike artists who bought ready prepared canvas from a colour shop), Beale kept to the established three-quarters and half-length sizes, also sometimes using a small half-length size, for most of his wife’s commercial work, largely reserving non-standard sizes to works painted privately or as gifts or favours.


Detail from reverse of Lemuel Abbott’s Viscount Bridport, 1785 (National Portrait Gallery). 75, in the compartment at right, give the canvas roll width in hundredths of a yard, thus 27 ins (68.6 cm). To the left, 17 is the roll length in yards, thus 17 yards (15.55 metres). Further left, 2224 is a progressive control number. The illegible small numerals, turned sideways at extreme right, give the year duty was levied. At left, the mark is partly covered by the wooden stretcher. Above, the other way up, is the stamp of the canvas supplier, James Poole.

Fig.2 Frame mark. Detail from reverse of Lemuel Abbott’s Viscount Bridport, 1785 (National Portrait Gallery). 75, in the compartment at right, give the canvas roll width in hundredths of a yard, thus 27 ins (68.6 cm). To the left, 17 is the roll length in yards, thus 17 yards (15.55 metres). Further left, 2224 is a progressive control number. The illegible small numerals, turned sideways at extreme right, give the year duty was levied. At left, the mark is partly covered by the wooden stretcher. Above, the other way up, is the stamp of the canvas supplier, James Poole.

Canvas was taxed in the 18th century. As a consequence of legislation in 1784, a 'frame mark', associated with the duty payable on linen, was applied by the Excise to the back of canvases. It can provide useful information both as to the width and the length of the roll from which the canvas was cut. [24] The mark took the form of a frame containing numbers and letters in compartments, as explained in the caption to fig.2. Hence on Lemuel Abbott’s portrait of Viscount Bridport, 1785, the canvas width is given as 75 hundredths of a yard, i.e. 27 ins, the canvas length as 17 yards and the year indistinct, while on George Morland’s Inside of a Stable, exhibited in 1791 (Tate), the canvas width is given as 165 hundredths of a yard, i.e. 59.4 ins, the canvas length as 265 (apparently 26 ½ yards) and the year as 1790. [25]

1.3 The availability of wide canvas

It is instructive to examine how canvas widths in paintings increased over the centuries. Narrower canvases, woven by a single person on a single loom, were more plentiful. Thus during the late 16th and early 17th centuries canvas was readily available in widths of between 40 and 50 ins (102 to 127 cm), as examination of whole-length and other large pictures at the National Portrait Gallery would confirm. [26] Canvas could be obtained in wider widths for specific needs, or could be made up by using a seam to join strips together.

Examples from the National Portrait Gallery include Marcus Gheeraerts’s Queen Elizabeth I of c.1592 at 60 ins wide without seam, Rowland Lockey’s The Family of Sir Thomas More of 1593 at 64 ins to the seam and Cornelius Johnson’s The Capel Family of c.1640 at 63 ins without seam. Van Dyck’s large equestrian Charles I of 1637-8 (National Gallery) is made up of two 72 ins pieces with a horizontal seam. [27] Such wide material was not necessarily readily available to painters in London, judging from the use of seams in later 17th century paintings in the National Portrait Gallery. In the 1660s Isaac Fuller’s five large paintings, Charles II’s escape after the Battle of Worcester, were painted on joined canvases, of up to 53 ins width to the seam, and at the end of the century John Closterman’s two large pictures, The Taylor Family (74¾ x 107 ins) and The Shaftesbury Brothers (95 ¾ x 67 ¼ ins) have seams at 56 and 59 ¾ ins respectively, approximately the width of a contemporary whole-length plus turnover (‘turnover’ indicates the canvas tacked to the side of the stretcher plus any loose canvas folded over the reverse).

It is worth looking at another centre of production, Haarlem, where the collection of the Frans Hals Museum has been the subject of detailed study. [28] Strip widths found in Haarlem in the 1590s and 1600s, typically 94.5 to 107 cm (37 ½ to 42 ins), are comparable to those employed in London. But from the 1610s to the 1680s, the Haarlem strip widths found in large paintings, typically 175 to 215 cm (69 to 84 ½ ins), generally exceed those found in London. In Haarlem the demand for large civic group portraits may have meant that there was a more consistent requirement among artists for wide canvases for prestige commissions, in turn making it worthwhile for specialist suppliers to source such products in a way that may not have been the case in London.

In a late 17th-century publication, The Merchant’s Ware-house laid open: or the Plain Dealing Linnen-Draper, published in 1695, all but one of the linens were of a width of a yard and a half or less (54 ins, 137 cm, or less). Exceptionally the broadest of the Irish Flaxens, at three yards wide (96 ins, 244 cm), recommended for sheets and table cloths, was described as ‘very fine and very strong… yet it is not commonly made use of’. [29]

In the 18th century, Sebastiano Ricci’s two large paintings on the staircase at the Royal Academy, The Triumph of Galatea and Diana and her nymphs bathing, both of the mid-1710s, have central sections of width about 67 ins (170 cm). This is about the width of the main section of Sir Joshua Reynolds’s pair of whole-length portraits in the Royal Collection, The Marquess of Granby and The Count of Schaumburg-Lippe. [30] George Stubbs’s Whistlejacket (National Gallery)of 1762 has a vertical seam at 72 ins. John Singleton Copley’s Death of Chatham (Tate, on loan to the National Portrait Gallery) of 1779-80 features a horizontal seam 71 ins up the 90 ins high composition. The production of extra-wide canvas was greatly facilitated by the invention of the flying shuttle in 1733 that allowed a single person to weave broadcloth. [31] Karl Anton Hickel’s William Pitt addressing the House of Commons (National Portrait Gallery) of 1793-5 is on a single canvas of size 127 x 177 ins (324 x 450 cm) while in 1798 George Romney took delivery from James Poole, his colourman, of ‘a cloth without seam on St[raining] fram[e] 13 f by 10 f’, that is 156 x 120 ins. [32]

From the late 18th century there was an increase in the width of canvas readily obtainable by artists in London, so much so that by the 1830s the colourmen, Roberson & Miller, could offer ‘Large Cloths, from 8 to 20 feet wide, Prepared on the Shortest Notice’, although how these were made up is not specified. [33] It was the introduction of power looms that facilitated the development of machine woven linen canvases in the second quarter of the 19th century. [34] By about 1905 Favor Ruhl in New York were advertising rolls at 18 x 36 ft for ceiling and wall decoration. [35] In about 1915, Henry M. Taws of Philadelphia illustrated a canvas of one piece, size 19 x 60 feet, made by E.H. & A.C. Friedrichs Company for the Museum of Natural History, New York. [36]

1.4 The range of standard canvas sizes

In the course of the 17th century, the size of portraits in all formats increased, whether head-and-shoulders, half-lengths or whole-lengths. We shall see how Cornelius Johnson upped his head-and-shoulders size in the 1620s and Peter Lely his half-length size in the 1650s and how the size of whole-lengths was also to increase by the end of the 17th century.

This increase in size probably reflects a number of inter-related influences: technically, through the ready availability of canvas and the greater flexibility it gave artists over panel; economically, through the rise of a richer society and with it the growth in the market for portraits; and culturally, through changing artistic and architectural taste which meant that interiors furnished with pictures became more fashionable. We shall also see how larger formats may have been introduced to meet specific artistic needs: the kit-cat size to allow an artist to include one or both hands without reverting to the much larger 50 x 40 ins size, the Bishop’s half length supposedly to accommodate the billowing sleeves of Hanoverian bishops and the Bishop’s whole length, or King size, as a response to Thomas Lawrence’s needs in painting portraits for a particularly grand space, the Waterloo Chamber at Windsor Castle. More generally, the rise in public exhibitions following the foundation of the Royal Academy in 1768 led to portraits being painted on a greater scale to command public attention.





Small half-length


Bishop’s half-length


Bishop’s whole-length

First used*








Early 19c

Size inches

24 x 20

30 x 25

36 x 28

44 x 34

50 x 40

56 x 44

94 x 58

106 x 70

Size ft ins

2 ft x 1ft 8

2 ft 6 x 2 ft 1

3 ft x 2 ft 4

3ft 8 x 2ft 10

4 ft 2 x 3 ft 4

4 ft 8 x 3 ft 8

7 ft 10x4 ft 10

8 ft 10 x 5 ft 10

Size cm

61 x 50.8

76.2 x 63.5


111.8 x 86.4

127 x 101.6

142.3 x 111.7

238.8 x 147.3

269.4 x 178











Fig.3 Standard sizes. The range of canvas sizes used by artists. First used*: approximate date when a particular size begins to be recognisable as a standard fixed size. Ratio**: canvas height divided by width.

The range of standard sizes used by artists and their development is shown in the table (fig.3).[37] The earliest sizes to be standardised, the three-quarters (30 x 25 ins, or 76.2 x 63.5 cm), which could accommodate a head-and-shoulders portrait cutting off the body at or above the waist, and the half-length (50 x 40 ins), usually a portrait to the knees, were in common use by the 1660s, if not before. Confusingly the modern terms generally used for these two sizes are the half-length and the three-quarter length respectively. At the end of the 17th century two further sizes came to be fixed, the kit-cat at 36 x 28 ins (91.5 x 71 cm) as an intermediate size between a three-quarters and a half-length, and the whole-length or full-length, which was usually 94 x 58 ins (249 x 147.3 cm). In a further elaboration in the late 18th century, the so-called Bishop’s half-length, usually 56 x 44 ins (142.3 x 111.7 cm) was introduced as a large half-length, followed in the early 19th century by the large whole-length, sometimes called a Bishop’s whole-length or King size, size 106 x 70 ins (269.4 x 152.5 cm). It was only in the mid-19th century that the system of standard sizes began to break down as portrait painters sought greater flexibility, and as other genres, not suited to portrait formats, became increasingly popular.

Were these particular standard sizes adopted for aesthetic reasons? In the second quarter of the 19th century, a French artist, Paillot de Montabert, writing about Parisian canvas sizes, claimed, ‘Artists generally believe that these dimensions have been determined by the experience of skilled masters, either for the good grace of portraits, or by some fitness of proportion which must be blindly respected: but it has nothing to do with all that’, and he concluded that the system was for the convenience of colourmen. [38] His arguments are acknowledged in two recently published studies of the French market, by Anthea Callen in 2000 and Pascal Labreuche in 2011. [39] Unlike in Antwerp where panel sizes were regulated in the early 17th century, there is no evidence that canvas sizes were a matter of legislation or guild regulation in Paris or in London. [40]

Looking at proportions, the height-to-width ratios for British standard sizes range between 1.2 and 1.29 for the smaller squarer sizes, to 1.62 and 1.51 for the largest sizes (see fig.3). In comparison, in France in the 18th century, the ratio for most canvases ranged between 1.2 and 1.33, increasing to 1.5 for the largest size. [41] While there is no evidence to support the idea that these proportions had any special significance to viewers, artists were of course aware of the proportions of canvas, as the testimony of James Northcote in the late 18th century shows. [42] But artists were content to keep to standard sizes for most purposes. And although the British whole-length at 94 x 58 ins closely matches the golden section, or golden number, of 1.618, again there is no evidence to suggest that this was the motivation for selecting these particular dimensions for this format.

The obvious reason for the standardisation of sizes in both Britain and abroad was convenience: convenience for the artist, for the artists’ supplier or colourman who stocked ready primed canvases, and for the framemaker. As portraits began to be produced more widely, the growing scale of production encouraged specialisation and standardisation. Primed cloth may have been readily available from specialist suppliers by 1668 when the anonymous author of The Excellency of the Pen and Pencil wrote that it could ‘be bought ready primed cheaper and better than you can do it your self. Few Painters (though all can do it) prime it themselves, but buy it ready done.’[43] There is some limited evidence that there were specialist primers at work in London even earlier in the century. [44] The rationale for standard sizes will be discussed further in section 5.2.

The availability of standard size canvases at fixed costs from London suppliers facilitated the production of portraits and the efficient operation of the studio since only with such standardisation could the system evolve which allowed suppliers to keep canvases and frames in stock to satisfy future demand. This was especially important to British portrait painting with its emphasis on producing a product at a relatively affordable price. A patron or artist in America, such as William Fitzhugh in 1698, John Smibert in 1749 or John Singleton Copley in 1771, could order canvases from England by name, three-quarters, kit-cats or half-lengths, without specifying the measurements and know with confidence what he would be getting. [45] Not only would standard canvases be kept in stock, so could appropriately sized picture frames. The London framemaker, Adrian Maskens, could thus advertise in 1780, ‘Three Quarters, Kit-Cats and Half-lengths may be had in a Minute’s Notice for Ready Money.’[46] As recently as 1959 in a trade publication on picture frames it was recommended, ‘In painting, paint regular sizes and have few but good frames (odd sizes are a rich man’s hobby)...’. [47]

In tracing the evolution of the range of standard sizes certain preliminary observations may be helpful. The terminology used by artists, colourmen and framemakers did not displace more obviously descriptive phraseology. In the inventory of Charles I’s collection drawn up in 1649-51, portraits are usually described as being ‘to ye knees’ or ‘at length’. [48] More than a century later Sir Joshua Reynolds used much the same terminology when writing to a prospective sitter in 1777: ‘My prices - for a head is thirty five Guineas - As far as the Knees seventy - and for a whole-length one hundred and fifty’. [49] A description such as a three-quarters would not necessarily have been understood by a client without explanation, although it was sufficiently accepted as a term to be used in the catalogues for public exhibitions held by the Society of Artists from 1760 and the Royal Academy from 1769.

Wright’s account book. Joseph Wright of Derby's record of sizes, contained on two pages of his account book, 1760s? (National Portrait Gallery).

Wright’s account book. Joseph Wright of Derby's record of sizes, contained on two pages of his account book, 1760s? (National Portrait Gallery).

Fig.4 Wright’s account book. Joseph Wright of Derby's record of sizes, contained on two pages of his account book, 1760s? (National Portrait Gallery).

Many a young portrait painter, colourman or framemaker must have noted down the historic standard sizes used by artists. However, the earliest known listings complete with measurements are those from the 1760s by Joshua Reynolds and Joseph Wright of Derby, both pupils of Thomas Hudson (see fig.4). [50] Before this period the evidence for the evolution of standard canvas sizes depends on observation of surviving portraits and a careful study of the limited documentation. Names like three-quarters, half-lengths and kit-cats are used in the late 17th and early 18th centuries in a manner which implies that their size was understood, whether in the notebooks of Charles Beale in 1677 and 1681, in correspondence concerning a leading artist, Sir John Medina in 1693, in orders for painting canvases from America in 1698, in the notebooks of George Vertue or in the account book of John Smibert in the 1720s and 1730s. [51]

Sizes are usually quoted in inches in this survey, with the equivalent in centimetres given where most useful, as well as in the table (fig.3). To have used the metric system throughout would have obscured the very subject of this study. For example, in a technical article on Joseph Wright of Derby’s Mr and Mrs Coltman (National Gallery), the picture was described as measuring 1.270 x 1.016 metres, [52] figures which are apparently accurate to the nearest millimetre yet obscure the fact that these measurements conform exactly to the standard canvas size of 50 x 40 ins.

Small deviations from standard sizes can be explained in various ways. In practice a canvas may sometimes have been stretched to a size slightly smaller or larger if the stretcher did not conform exactly to standard measurements. More commonly a canvas may have lost its margin in relining or may have been given a new stretcher of slightly different size during conservation. Sometimes paintings were measured in the frame (the ‘sight size’), as acknowledged to be the case in the catalogues of the National Portrait Gallery until 1914. One way or another the result is that a nominal 30 x 25 ins canvas may be found in catalogues with marginally differing measurements, most commonly 29 ½ x 24 ½ ins. In any case, some artists appear to have adopted a slightly different standard on occasion. [53]

When a portrait does not conform to a standard size, it is worth asking why. Indeed, as Desmond Shawe-Taylor has suggested, ‘Any canvas not of these dimensions has been cut or is of a special non-standard size sometimes used for group portraits’. [54] Unless an artist is known to have used other sizes, a non-standard canvas for a life-size portrait would suggest that it may have been cut down, enlarged or painted for a special purpose. If an explanation can be found it may illuminate the history of the picture. An example is Michael Dahl’s whole-length Admiral Sir Cloudesley Shovell, 1702 (National Portrait Gallery) which at 87 ½ ins stands out as being shorter than the standard 94 inches of a whole-length. Studio examination revealed that the top four or five inches of canvas were folded over many years ago, perhaps even in the mid-18th century, judging from the style of the frame. [55]

Found a mistake? Have some extra information? Please contact Jacob Simon at [email protected].


[1] Thomas Page, The Art of Painting, Norwich, 1720, pp.43-4; the reference to a ‘three quarter’d’ is unusual and recurs on p.126, where a ‘Three quartered cloth’ is referred to. This style can be found in a rather different context in 1698 when an order was sent to London for 'Six three quartered lacker book frames for pictures well burnished, see Richard Beale Davis (ed.), William Fitzhugh and his Chesaspeake World, 1676-1701, Chapel Hill, 1963, p.367.

[2] In the most complete survey of the period, Roy Strong, The English Icon: Elizabethan and Jacobean portraiture, 1969, about 5% of 16th-century works are recorded as being on canvas. Before the 1570s surviving works on canvas are quite exceptional. In the early 17th century canvas came into common use with the work of artists such as Robert Peake the Elder, John De Critz, Marcus Gheeraerts and William Larkin. For the use and supply of linen as a support at the beginning of the period under review in this survey, see Maria Hayward, ‘The London Linen Trade, 1509-1641, and the use of linen by painters in royal service’, in Jo Kirby et al. (eds), Trade in Artists' Materials: markets and commerce in Europe to 1700, 2010, pp.375-85.

[3] The statement that the Ditchley portrait was painted on a single canvas is based on personal observation and is confirmed by x-ray photographs. The sonnet on the right in the portrait appears to have been cut through, leading Roy Strong to conclude that the picture had been cut down ‘on either side by at least three inches’ (Roy Strong, Tudor and Jacobean Portraits, 1969, p.107). If so, then the original piece of canvas may have been as wide as 68 ins (173 cm), allowing for tacking edges.

[4] Two essays touching on the standardisation of sizes in England, by Caroline Rae and Ian Tyers, are in preparation for the forthcoming publication, Tarnya Cooper et al. (eds), Painting in Britain 1500-1630: production, influences and patronage, planned for 2015.

[5] The word ‘canvas’ derives from ‘cannabis’, the Latin word for hemp. Linen was the normal canvas used for portrait painting in Britain in the 18th and 19th centuries, often available in various grades. The range of canvases advertised by artists’ colourmen increased during the late 19th and early 20th centuries. In January 1910 Reeves & Sons were offering various flax canvases but also cotton and jute (see Price List of Artists’ Materials, pp.146-7). For a summary history, see Caroline Villers, ‘Artists Canvas: a history’, in ICOM Committee for Conservation, 6th triennial meeting, Ottawa, 1981: preprints, Paris, 1981, p.81/2/1/8.

[6] The Artist’s Repository and Drawing Magazine, n.d. but 1785, vol.3, p.63.

[7] Theodore Sizer (ed.), The Autobiography of Colonel John Trumbull, Patriot-Artist, 1756-1843, New Haven, 1953, p.278. On occasion, household linens were used as a supports by artists; for a short survey of continental practice, see Joan Marie Reifsnyder, ‘Artist's Canvas – Or Just an Old Tablecloth?’, in Painting Techniques. History, materials and studio practice, summaries of the posters at the 1998 International Institute for Conservation Congress, Dublin, unpaginated.

[8] Daniel Defoe, A Plan of the English Commerce, 1728, pp.78, 207.

[9] J.F., The Merchant’s Ware-house laid open: or, the plain dealing linnen-draper, 1695.

[10] See N.B. Harte, ‘The Rise of Protection and the English Linen Trade, 1690-1790’, in N.B. Harte and K.G. Ponting (eds), Textile History and Economic History: Essays in honour of Miss Julia de Lacy Mann, Manchester, 1973, especially pp.74-85.

[11] Ibid, pp.108-9.

[12] See Mansfield Kirby Talley, Portrait Painting in England: Studies in the technical literature before 1700, 1981, p.283. It is possible that the Gentish Holland was for Mary Beale’s personal use rather than for painting.

[13] Plain cloth, Roman and ticken feature regularly in 19th-century manuals and catalogues. Roman appears in Roberson & Co’s catalogue of c.1840-53 (Roberson Archive, Hamilton Kerr Institute, Cambridge, information from Sally Woodcock) and in later catalogues of artists’ materials. It is a rep (double weft yarns alternating with single warp yarns); see Rosamond D. Harley, ‘Artists’ Prepared Canvases from Winsor & Newton 1928-1951’, Studies in Conservation, vol.32, 1987, p.80. Ticken is of much older origin as a support for paintings and is a canvas with a twill weave. Bed-ticking features in the notebooks of Charles Beale in 1677 (Talley, op.cit., p.283). In The Artist’s Repository in 1785, op.cit. in note 6, specific reference is made in the entry for picture cloths to ‘A Sort of Ticking...made by Mr. W. Middleton of St. Martin’s Lane’. Roberson & Miller were advertising ‘Prepared Cloths and Tickens’ in their single-sided sheet, Materials for Drawing and Painting, c.1828-40, and similar references occur in most later 19th-century catalogues.

[14] R.D. Connor, The Weights and Measures of England, 1987, p.94.

[15] Roberson & Miller, Materials for Drawing and Painting, single-sided sheet, c.1828-40.

[16] Connor, op.cit., p.94.

[17] Favor, Ruhl & Co, Trade Price List of Artists’ Materials, n.d. but c.1905, p.35. The catalogue does not make clear whether the the extra two inches reflected the actual rather than nominal measurements of the canvas or came about as part of the stretching process itself.

[18] Connor, op.cit., pp.94-6. The use of the ell in measuring linen is identified by Lewis Roberts, The Merchants Mappe of Commerce, 1638, as cited in Jo Kirby, ‘The Painter’s Trade in the Seventeenth Century: Theory and Practice’, National Gallery Technical Bulletin, vol.20, 1999, p.22. In England the ell was 45 ins, in Scotland 37 ins, while the Flemish ell was 27 ins. In selling cloth it was customary to place the thumb beyond the end of the measure, and cut across the cloth just beyond the thumb, thus giving an extra inch, so that if the cut were not quite straight the buyer would not lose. In time it became customary to give an extra inch for each yard, not just an extra inch in the total length of cloth as cut; see Connor, pp.87, 93-4.

[19] See Christopher Brown and Nigel Ramsay, ‘Van Dyck’s collection: some new documents’, Burlington Magazine, vol.132, 1990, p.708, and Robin Gibson, Catalogue of Portraits in the Collection of the Earl of Clarendon, 1977, p.139.

[20] Bodleian Library, University of Oxford, Charles Beale’s diary notebook for 1677, entry for 20 February.

[21] George Vertue, ‘Vertue Note Books’, vol.5, p.20, published as Walpole Society, vol.26, 1938; see also ‘Vertue Note Books. Index to volumes I-V’, vol.6, p.122, published as Walpole Society, vol.29, 1947.

[22] Jabez Barnard, Price Catalogue of Materials, in Edwin Jewitt, Manual of Illuminated and Missal Painting, c.1860. Hatchments were commonly size 45 x 45 ins, or an ell square.

[23] The 1677 notebook is in the Bodleian Library and that for 1681 in the National Portrait Gallery Archive. The fullest account of their contents is given in Kirby Talley, op.cit. in note 12. Other notebooks in the same series were known to George Vertue but are now lost.

[24] The best informed account of the history of the frame mark and its meaning is H. Dagnall’s The Marking of Textiles for Excise and Customs Duty, Edgware, 1996, pp.15-18, 26-7. Markings have been found, among others, of 70, 75 and 80 hundredths of a yard; the supposition made here is that cloth had to be at least as wide as the indicated width to be so marked. On this basis the actual width may in some cases have been slightly greater than the indicated or nominal width.

[25] The Morland duty frame mark is reproduced by Martin Butlin, ‘Turner’s Late Unfinished Oils: some new evidence for their late date’, Turner Studies, vol.1, no.2, 1981, p.44.

[26] The analysis of canvas widths at the National Portrait Gallery is based on an examination of ten whole-length portraits on canvas from 1590 until 1620. In date order (by unknown artist unless stated): 1st Baron Burghley, 1590s, 38 ½ ins (NPG 4881), Marcus Gheeraerts the Younger’s 2nd Earl of Essex, c.1597, 50 ins (NPG 4985), 3rd Earl of Southampton, c.1600, 48 ins (NPG L114), 1st Earl of Nottingham, (1602), 48 ins (NPG 4434), Sir Walter Ralegh and his son, 1602, 50 ins (NPG 3914), The Somerset House Conference, 1604, 43 ½ ins (NPG 665), John De Critz the Elder’s Anne of Denmark, c.1605-10, 49 3/4 ins (NPG 6918), 4th Earl of Pembroke, c.1615, 40 ins (NPG 5187), attributed to William Larkin, 1st Duke of Buckingham, c.1616, 40 ins (NPG 3840), attributed to Paul van Somer, Monmouth family, c.1617, 48 ½ ins (NPG 5246), attributed to Abraham van Blyenberch, King Charles I, c.1617-20, 41 ins (NPG 1112).

[27] Kirby, op.cit. in note 18, p.23.

[28] Ella Hendriks, ‘Canvas supports’, in Pieter Biesboer et al., Painting in Haarlem 1500-1850: the collection of the Frans Hals Museum, 2006, pp.70-4, 86-7. The author tentatively suggests that artists working on large paintings in Haarlem were more likely to use wide canvas than those in nearby Amsterdam, who more often used joined strips.

[29] J.F., op.cit. in note 9, p.15.

[30] Oliver Millar, The Later Georgian Pictures in the Collection of Her Majesty the Queen, 1969, nos 1022, 1027, describes the Reynolds as enlarged on the right in painting; however, the visual evidence would suggest that these portraits were not enlarged in painting but painted on canvases joined at the outset.

[31] Connor, op.cit. in note 14, p.91.

[32] A late 19th-century note in the Hickel file records what appears to be a duty frame mark, visible before relining, including the numerals, 351, which would indicate a canvas width of 126 ins, very close to the actual canvas width (National Portrait Gallery, RP 745). For Romney, see Mary Bustin, ‘Mrs Robert Trotter of Bush (1788-9)’, Transactions of the Romney Society, vol.2, 1997, p.9.

[33] Roberson & Miller, op.cit. in note 15.

[34] Villers, op.cit. in note 5, p.81/2/1/8.

[35] Favor, Ruhl & Co, Trade Price List of Artists’ Materials, n.d. but c.1905, p.36.

[36] Henry M. Taws, Catalog of Artists and Draughtsmens Materials, n.d. but c.1915, p.23. .

[37] Prior to the standardisation discussed in this survey, certain sizes found favour, at least within the confines of an individual studio and sometimes more widely; see the forthcoming essay by Caroline Rae, referred to in note 4.

[38] Jacques-Nicholas Paillot de Montabert, Traité complet de la peinture, Paris, 1829-51, vol.9, p.147, as quoted by Anthea Callen, The Art of Impressionism: painting technique & the making of modernity, New Haven and London, 2000, p.18.

[39] Callen, op.cit., pp.18-19; Pascal Labreuche, Paris, capitale de la toile à peindre, XVIIIe-XIXe siècle, Paris, 2011, pp.36-40.

[40] For Antwerp, see Jørgen Wadum, ‘The Antwerp Brand on Paintings on Panel’, in Erma Hermens (ed.), Looking Through Paintings, Leids Kunsthistorisch Jaarboek, vol.11, 1998, p.182. For Paris, see Labreuche, op.cit., p.36.

[41] For the proportions found in French canvases, see Labreuche, op.cit., pp.37-40. For the purposes of the present survey, the dimensions given by Pernety in 1757 have been used, rather than those published by Savary de Bruslons in 1730.

[42] James Northcote used one of the preliminary pages in his account book to explore questions around proportion (see Jacob Simon, ‘The Account Book of James Northcote’, Walpole Society, vol.58, 1996, p.33). There he noted Benjamin West’s belief that 5 x 7 feet was ‘a good proportion’ for the artist’s The Battle of La Hogue (c.1778). He recorded various canvas sizes as used by Van Dyck, Hogarth, Reynolds and himself, describing 2 feet 11 inches by 29 inches (35 x 29 ins) as ‘best size’. A few years later on 6 September 1794, the artist Joseph Farington wrote of a work by Sir George Beaumont, ‘The proportion of Sir George's Canvass recommended by Mr. West is 4 feet 8 inches by 3 feet 4, and I think it a very agreeable proportion’ (K. Garlick and A. MacIntyre (eds), The Diary of Joseph Farington, vol.1, 1978, p.231).

[43] Anonymous, The Excellency of the Pen and Pencil, 1668, p.94.

[44] For an ‘imprimeur Wallon’ in London, mentioned by Theodore de Mayerne, apparently in the 1620s, see M. Faidutti et C. Versini (eds), Pictoria sculptoria & quae subalternarum artium. Le manuscrit de Turquet de Mayerne, Lyon, 1970, pp.14-7. For ‘Fenn the Liegois’, who primed canvas for Robert Walker, according to Richard Symonds writing in 1652, see Mary Beal, A Study of Richard Symonds: his Italian notebooks and their relevance to 17th century painting techniques, PhD thesis, University of London, 1978, University Microfilms International, 1980, pp.307, 311.

[45] For Fitzhugh, see Richard Beale Davis (ed.), William Fitzhugh and his Chesaspeake World, 1676-1701, Chapel Hill, 1963, p.367, for his order for ‘half a doz. 3 quarter clothes to set up a painter'. For Smibert, see Richard H. Saunders, John Smibert. Colonial America’s first portrait painter, New Haven and London, 1995, pp.257-9, for his orders to his fellow-artist in London, Arthur Pond, e.g. on 6 April 1749 for ‘¾ Primed Cloaths 3 doz. Kit Kats 1 doz. ½ Lengths 2 doz.’. For Copley, see Letters & Papers of John Singleton Copley and Henry Pelham 1739-1776, New York, 1970, p.115, for an order sent to London for him in 1771 by his relative, Henry Pelham, for ‘12 half Length Cloths. 6 kitkat D[itt]o. 12 ¾ D[itt]o.’ for delivery to Boston.

[46] For Maskens, see British picture framemakers on the National Portrait Gallery website.

[47] Armand Blackley, writing in the catalogue of James Bourlet, 1959.

[48] Oliver Millar (ed.), ‘The Inventories and Valuations of the King’s Goods, 1649-1651’, Walpole Society, vol.43, 1972, pp.70-71, etc.

[49] John Ingamells and John Edgcumbe (eds.), The Letters of Sir Joshua Reynolds, New Haven and London, 2000, p.69.

[50] For Reynolds, see the listings in his pocket books for 1760, 1764, 1765, 1766, 1770, 1771 and 1782 in the Royal Academy Library. For Wright, see Elizabeth Barker, ‘Documents relating to Joseph Wright of Derby (1734-97)’, Walpole Society, vol.71, 2009, pp.46, 50.

[51] These sources are quoted elsewhere in the text of this survey except for the reference to Medina using half-lengths and three-quarters, which is to be found in John Fleming, ‘Sir John Medina and his “Postures”’, Connoisseur, vol.148, 1961, pp.22-5.

[52] Martin Wyld and David Thomas, ‘Wright of Derby’s “Mr and Mrs Coltman”: An Unlined English Painting’, National Gallery Technical Bulletin, vol.10, 1986, p.29.

[53] What evidence there is on variations in sizes relates to Jonathan Richardson, Joseph Wright of Derby, Thomas Gainsborough, George Romney and George Barret. Richardson chose to use an extra-large half-length size late in life (see note 62). Wright quotes the height of a kit-cat as 361/4 ins, not 36 ins (see note 50). Both Gainsborough and Romney varied the sizes of their whole-lengths on occasion, as will be discussed. Barret quotes the height of a three-quarters as 29 ins, rather than 30 ins (see Anne Crookshank and the Knight of Glin, The Painters of Ireland c.1660-1920, 1978, p.117).

[54] Desmond Shawe-Taylor, The Georgians: Eighteenth-century portraiture & society, 1990, p.10.

[55] Memorandum by Jacob Simon, August 1994 (Shovell file, National Portrait Gallery, RP 797).