A justification for this study of Veronese's paintings in the National Gallery is that the collection includes examples of both his early and late work and also examples of the different types of painting that he undertook. But some indication of the limitations of the sample should be noted. No examples of canvases with relatively small figures have been included. These are not only some of the most exquisite of Veronese's paintings, but most scholars agree that some of them are likely to date from late in his career. Veronese died in 1588 and was certainly active during the 1580s – for example the great altarpiece now in the Musée des Beaux Arts, Dijon, was commissioned in 1586(note 80)and the high altar of San Pantaleone in the following year (note 81) – none of the paintings in the National Gallery has been dated so late.
One of the most remarkable of the smaller late paintings by Veronese is the beautiful signed ‘Crucifixion’ (Fig. 8) painted on stone, which is in the Museo Civico, Padua (note 82). Here, as in other (later) examples by artists from Verona, much of the picture surface consists simply of the highly polished stone which represents the deep black of the night sky (note 83). That in itself is of interest, but so, too, is the idea of painting on a dark ground which this technique necessarily involves. The latter part of Veronese's career coincided with a period of change in painting technique in Italy – experimentation with alternative supports and dark grounds were part of this change. Some of Veronese's late canvas paintings of the Passion, most notably ‘The Agony in the Garden’ in the Brera Gallery in Milan (note 84), clearly have such a ground. Nor did the artist favour it only for sombre subjects. One of the most beautiful of his late small canvases, ‘The Finding of Moses’ in the Musée des Beaux-Arts, Lyon, is also painted on a dark ground (note 85).
The larger painting of ‘Mars and Venus’ (Fig. 9) in the National Gallery of Scotland, generally regarded as a late work, and by some as a workshop production, is painted on a dark ground but with a dense area of lead white for the radiant nude body of the goddess (Fig. 10), with an effect more dramatic today than the artist is likely to have foreseen (note 86).
However routine Veronese's followers and assistants were as artists, they seem to have participated in the experiments of the artist's last years. Striking testimony to their interest in technical innovation is the large altarpiece of the ‘Pietà and Saints’ the family workshop is known to have supplied to S. Giobbe in Venice which is painted on metal, probably copper – one of the very few altarpieces anywhere painted on such a support (note 87).
Even if there are aspects of Veronese's technique which are not represented by the paintings in the National Gallery examined in this article, we can feel sure that the works discussed here are likely to be characteristic of the mainstream of Veronese's work, and, indeed, central to Venetian practice of painting on canvas in general. The National Gallery's Veroneses thus provide norms against which other examples of his painting – and that by other artists working in Venice and the Veneto in the mid-sixteenth century – can be measured.
Notes and References
1. For the inventory taken after 15 February 1637 see C. Gould, ‘The Sixteenth Century Italian Schools’, National Gallery Catalogues, London 1959 (revised in 1975), p. 329.
2. For the earlier inventory of 1621 – also posthumous because Rudolph died in 1612 – see Gould, cited in note 1, p. 328. Gould regarded it as merely 'a possibility that they had been commissioned by Rudolf. It is surely a strong probability, for no other pictures of this character and quality seem to have entered the Imperial Collection between Rudolph's death and 1637. The schemes employed on Venetian ceilings are discussed by J. Schulz in ‘Venetian Painted Ceilings of the Renaissance’, Berkeley and Los Angeles 1968. It may not be generally accepted that the National Gallery's ‘Allegories’ were intended for a ceiling – Gould observed merely that it was 'plausible enough' (p. 328).
3. Full discussions of this drawing are to be found in R. Cocke, ‘Veronese's Drawings’, London 1984, pp. 184–5, no. 78, and in W. R. Rearick, ‘The Art of Paolo Veronese’, exhibition catalogue, National Gallery of Art, Washington 1988, no. 62, pp. 125–6.
4. For a brief general survey of Veronese's work for Rudolph see Rearick, cited in note 3, pp. 120–2.
5. Rudolph's marriage to Isabella, eldest daughter of Philip II of Spain (1566–1633), was planned in 1571 but came to nothing. His liaison with Anna Maria Strada, by whom he had six children, commenced in 1583 and it is just possible that the union celebrated in these pictures was with her. That might explain some aspects of the subject matter, such as the emphasis on lust, even lust overcome, and the episode of the woman (apparently) shared by two men (Anna Maria was married) which might seem surprising for a bride. And then if these pictures were in a residence of the Emperor's mistress the interval before they were inventoried as part of the Imperial Collection might also be explained. Speculations as to the meanings of the ‘Allegories’ will be found in E. Wind, ‘Pagan Mysteries in the Renaissance’, Harmondsworth 1967, appendix 8, pp. 272–5; A. Braham, 'Veronese's Allegories of Love', ‘Burlington Magazine’, 1970, pp. 158–62; M. Royalton-Kisch, 'A New Arrangement for Veronese's Allegories of Love in the National Gallery', ‘Burlington Magazine’, 1978, pp. 158–62; Rearick, cited in note 3, pp. 126–9.
6. Royalton-Kisch (cited in note 5) first raised the issue of the figures looking out, proposing that they were looking from one picture to another. The suggestion about the heraldic centrepiece is Rearick's (cited in note 3, p. 129). He seems not to have realised that Rudolph never married. Royalton-Kisch was the first to propose in print an arrangement like that found in our figure 1, although his arrangement was slightly different.
7. Gould, cited in note 1, p. 328.
8. Ibid. pp. 328–9 and note 3 on p. 330 citing the catalogue of 1929 and the catalogue for the 1947 exhibition of cleaned pictures.
9. Veronese's ‘Allegories’ had been last cleaned and restored between 1946 and 1950: NG 1318 and 1326 (1946); NG 1324 (1950); NG 1325 (1951).
10. Principally examination of paint cross-sections and thin sections by optical microscopy in reflected and transmitted light.
11. Spectrographic (elemental) analysis with the laser microprobe, now superseded by SEM-EDX analysis. For the earlier method, see A. Roy, 'The Laser Microspectral Analysis of Paint', ‘National Gallery Technical Bulletin’, 3, 1979, pp. 43–50.
12. For work on green copper-containing glazes, see, for example, J. Pilc and R. White, 'The Application of FTIR-Microscopy to the Analysis of Paint Binders in Easel Paintings', ‘National Gallery Technical Bulletin’, 16, 1995, p. 78 and p. 82.
13. When taken out of their frames for conservation treatment, thin discontinuous painted borders (c. 2cm wide) of red-brown and orange-brown were discovered, roughly marking out the edges on each of the compositions. They were most probably drawn on to the finished paintings as part of a procedure to size the pictures for their eventual architectural settings, since Veronese may well have had to make these calculations at a distance. Cross-sections from these borders and analysis of the materials, including the identification of orpiment and realgar, suggest that they are contemporary with Veronese's paintings, but there is also evidence for reinforcement of these painted outlines in red-brown earth pigment passing over old varnish, representing perhaps minor readjustments to the sizes made during the installation of the pictures, or in preparation for framing as later independent compositions.
14. J. Plesters, A. Roy and D. Bomford, 'Interpretation of the magnified image of paint surfaces and samples in terms of condition and appearance of the picture', in N. S. Brommelle and G. Thomson (eds. ), ‘Science and Technology in the Service of Conservation’, Preprints of the IIC Washington Congress, London 1982, pp. 169–70.
15. Thin cross-sections show fading of red lake dyestuffs in the upper portion of surface glazes; also glaze paints protected from light on turnover edges retain a stronger colour than adjacent exposed paint.
16. See Rearick, cited in note 3, p. 128.
17. A clear example of smalt particles retaining a blue core in a sample from Veronese's ‘Consecration of Saint Nicholas’ (NG 26) is published in the first part of this article. See N. Penny and M. Spring, 'Veronese's Paintings in the National Gallery. Technique and Materials: Part I', ‘National Gallery Technical Bulletin’, 16, 1995, p. 13, plate 8. For the mechanism of discoloration, see B. Mühlethaler and J. Thissen, 'Smalt' in ‘Artists' Pigments. A Handbook of Their History and Characteristics’, Vol. 2, ed. A. Roy, 1993, pp. 116–21.
18. Plesters, Roy and Bomford, cited in note 14. For a thin section of browned copper green glaze from 'Respect', see H. Kühn, 'Verdigris and Copper Resinate' in A. Roy (ed.), cited in note 17, p. 152.
19. For the essentials of Venetian canvas painting technique, see A. Lucas and J. Plesters, 'Titian's "Bacchus and Ariadne'", ‘National Gallery Technical Bulletin, 2’, 1978, pp. 25–47; J. Plesters, 'Tintoretto's Paintings in the National Gallery. Part II, Materials and Techniques', ‘National Gallery Technical Bulletin’, 4, 1980, pp. 32–47; also, L. Lazzarini, 'Il Colore nei Pittori Veneziani tra il 1480 e il 1580', ‘Bollettino d'Arte, Studi Veneziani, Ricerche di Archivio e di Laboratorio’, Supplemento 5, 1983, pp. 135–44.
22. No complete X-ray mosaics were made of the four ‘Allegories’, although parts of each were X-rayed in 1981–2.
23. See Part I of this article, Penny and Spring, cited in note 17.
24. For NG 1324, see J. Mills and R. White, 'Analyses of Paint Media', ‘National Gallery Technical Bulletin’, 5, 1981, pp. 66–7; for NG 1318, 1325 and 1326, see J. Mills and R. White, 'Analyses of Paint Media', ‘National Gallery Technical Bulletin’, 7, 1983, pp. 66–7.
25. Smalt was also detected in the sky and elsewhere in ‘The Family of Darius before Alexander’ (NG 294), evidently an important commission. See N. Penny and M. Spring, cited in note 17, pp. 16–22. Similarly, a commission from the Holy Roman Emperor, for the ‘Allegories’, would have been significant in value and status.
26. See J. Mills and R. White, 'Organic Mass-Spectrometry of Art Materials: Work in Progress', ‘National Gallery Technical Bulletin’, 6, 1982, pp. 10–13.
27. Plesters, Roy and Bomford, cited in note 14.
28. The connection of lead-tin yellow (II) to Venice in the sixteenth century and the occurrence of both varieties of the pigment is commented on by H. Kühn in 'Lead-Tin Yellow', A. Roy (ed. ), cited in note 17, pp. 85–9 and p. 101.
29. Ibid., p. 85 and pp. 95–7.
30. E. Martin and A. Duval, 'Les deux variétés de jaune de plomb et d'étain: étude chronologique', ‘Studies in Conservation’, 35, 1990, pp. 117–25.
31. Kühn, cited in note 28, pp. 95–6.
32. Verdigris particle types are sometimes recognisable in paint cross-sections, particularly thin sections, by comparison with scanning electron micrographs of standard specimens. In the greens from the ‘Allegories’, blue basic verdigris ([Cu(Ac)2]2. Cu(OH)2. 5H2O) and green basic verdigris (Cu(Ac)2. [Cu(OH)2]3. 2H2O) were identified. See Kühn, 'Verdigris and Copper Resinate', A. Roy (ed. ), cited in note 17, pp. 134–5, fig. 5.
33. The chloride is present at relatively low levels and seems not to indicate the use of an artificial basic copper chloride pigment such as atacamite or calumetite. More likely it is a residual component from one of the traditional methods for preparing verdigris, involving exposing to acetic acid vapour copper plates coated in common salt (sodium chloride) bound with honey. See, for example, recipe no. 85, 'to make verdigris', in Mrs Merrifield's translation of the Bolognese MS, in ‘Original Treatises on the Arts of Painting’, Vol. 11, 1849, p. 418.
34. Mills and White, cited in note 26.
35. Natural arsenic(II) sulphide can exist in several polymorphs. A new form, pararealgar, was identified in 1980 and the first occurrence in a painting was reported in 1995. See M.-C. Corbeil and K. Helwig, 'An occurrence of pararealgar as an original or altered artists' pigment', ‘Studies in Conservation’, 40, 1995, pp. 133–8. These authors point out that the pararealgar could have resulted from the transformation of realgar.
36. T. Pignatti, ‘Veronese’, 2 vols., Venice 1976, I, no. 256, p. 150; R. Pallucchini, ‘Veronese’, Milan 1984, no. 232, p. 232 and p. 186; Rearick, cited in note 3, no. 71, p. 139.
37. Pignatti, cited in note 36, I, nos. 84–90, 120–2, pp. 116, 124; II, figs. 176–84, 362–4.
38. K. Brugnolo Meloncelli, ‘Battista Zelotti’, Milan 1992, cat. 21, pp. 101–2; figs. 122–3.
39. Ibid. cat. 18, p. 100; figs. 91–4.
40. For the engraving by a follower of Marcantonio see Bartsch, ‘Le Peintre Graveur’, XIV, no. 460 (‘The Illustrated Bartsch’), XXVII (ed. K. Oberhuber), New York 1978, no. 460 (342). Chiaroscuro prints after the Marcantonio were made by Zanetti in the eighteenth century. It is sometimes claimed (e. g. by Rearick, cited in note 3, p. 139) that Veronese's source was a chiaroscuro print and specifically one by Ugo da Carpi. There may be a confusion between Ugo da Carpi's print of a sibyl after Raphael and the Zanetti (see ‘The Illustrated Bartsch’, XLVIII (ed. C. Karpinski), New York 1983, pp. 138–9, 340–1). The drawing upon which Marcantonio's print is based is certainly among the most Parmigianesque of Raphael's drawings, and was generally regarded as by Parmigianino at the time of the first edition of Gould's catalogue, but no authority has doubted the attribution to Raphael since the brilliant article by Konrad Oberhuber, 'Eine unbekannte Zeichnung Raffaels in den Uffizien', ‘Mitteilungen des Kunsthistorischen Institutes in Florenz’, XII, 1966, pp. 225–44.
41. Gould, cited in note 1, p. 325.
42. Carlo Ridolfi, ‘Le Maraviglie dell'Arte’, 2 vols., ed. D. von Hadeln, Berlin 1914, I, p. 310, note 3.
43. For example Pignatti, cited in note 36, I, no. 206, p. 141, expressing agreement with Pallucchini and observing the 'sontuosità raffinatissima'. Rearick, cited in note 3, p. 139, dates the picture 'about 1558' but gives no reasons for doing so.
44. The painting measures 197.5 x 115.6 cm. The narrow strip (c. 15 cm not including the tacking edge) along the left edge is joined by a vertical seam. The larger piece is therefore almost exactly a metre in width but, with the tacking edge, would originally have measured a little more than a metre. The thread count, measured from the X-ray, is on average 10 x 10 threads per cm.
45. CaSO4. 2H2O identified by X-ray diffraction (in agreement with JCPDS file no. 6–46).
46. ‘Gesso sottile’ is rehydrated calcium sulphate, prepared by slaking burnt gypsum. Raw gypsum is also calcium sulphate dihydrate, and therefore indistinguishable from ‘gesso sottile’ by X-ray diffraction but, given the coarse texture of raw gypsum, it seems more likely that the ground here is ‘gesso sottile’.
47. E. Martin, N. Sonoda, A. R. Duval, 'Contribution à l'étude des préparations blanches des tableaux italiens sur bois', ‘Studies in Conservation’, 37, 1992, pp. 82–92.
48. Smalt, a blue cobalt-containing glass, was identified by EDX.
49. Many of the particles show a loss of colour around their edges, only the core remaining blue. See Mühlethaler and Thissen, 'Smalt', in A. Roy (ed. ), cited in note 17, pp. 116–21.
50. Vermilion, and the alumina substrate used in the preparation of the red lake pigment, were identified by EDX analysis on a cross-section.
51. EDX analysis indicated that the bright yellow pigment in paint from the orange drapery contained As. The appearance under the microscope was characteristic of orpiment (arsenic[III] sulphide), rather than the yellow polymorph of arsenic(II) sulphide pararealgar (see note 35).
52. For example, G. P. Lomazzo, ‘Trattato dell'arte de la pintura’, Milan 1584, Libro Terzo: Del Colore, Cap. VI, p. 193. 'L'oro pimento è nemico di tutti i colori, saluo che del giesso, ocrea, azurri, smalti, verdi azurri, terra verde, morel di ferro, endico, maiolica, e lacca'. (Orpiment is an enemy to all save gypsum, ochre, azures, smalt, green azure, green earth, rust of iron, brown of Spain and lake.)
54. See Kühn, cited in note 28, pp. 99–111. Results of analyses carried out by X-ray diffraction at the National Gallery are published in this article, together with Kühn's results; the reported occurrences of lead-tin yellow 'type II' in the sixteenth century are all in Venetian paintings.
55. Kühn, cited in note 28, p. 91. Kühn notes that although both types of lead-tin yellow are usually of small particle size, larger particles sometimes occur in samples of the 'type IF form.
56. First noted by F. Sansovino, ‘Venetia città noblis-sima et singolare’, Venice 1581, fol. 65r. Neither this nor any subsequent Venetian guide describes it as an altarpiece. That it was on the left of the church is clear from the ‘Descrizione’ of Boschini and Zanetti in 1733 (p. 269) and there is no reason to suppose that it had moved between then and 1821 when A. Quadri in ‘Otto Giorni a Venezia’ (I, 1821, p. 273) describes it as hanging ‘between’ the second and third altars. Interestingly, Eastlake himself never seems to have described it as an altarpiece. Moreover, it is considerably wider than any other painting by Veronese or Tintoretto that is known to have hung above an altar.
57. ‘Annual Report of the National Gallery for 1856’, p. 27, and MS Minutes of the Board, IV (1855–71), pp. 6–8 (for Eastlake's letter of 12 November 1855 proposing the purchase) and pp. 17–19 (for his report of 9 February 1856).
58. Minutes of the Board, cited in note 57 above. In October 1855 the painting was packed up for transport to Paris where it was to be offered to Baron James Rothschild. Otto Mundler engineered its diversion, but, even after the sale of the painting, a higher offer was (according to Eastlake) made by a person who had 'before endeavoured to come to an arrangement' with the vendor (probably an allusion to the Baron).
59. Gould, cited in note 1, pp. 318–19.
60. Gould, cited in note 1, p. 319.
61. Richter's report of Morelli's opinion is given in a letter to A. M. Daniel of 8 January 1932 (National Gallery Archives). Waterhouse had a few years previously convinced himself that none of the painting was by Veronese (his note in the Gallery's dossier).
62. Pignatti, cited in note 36, I, no. 231, pp. 144–5; II, figs. 545 and 546.
63. Pen and brown ink heightened with white on pale (faded) blue paper, 27.8 x 20.2 cm. Teylers Museum, Haarlem, B65. For a full discussion of the drawing see Rearick, cited in note 3, no. 58, pp. 115–16. He, however, makes no mention of the altarpiece in S. Corona which, as Gould correctly notes, also owes something to ideas sketched out here.
64. The standard loom width for paintings in Venice, from measurements on paintings, appears to be just over one metre.
65. EDX analysis on cross-sections showed that there were some large particles containing Al, Si and K.
66. R. J. Gettens, E. West Fitzhugh and R. L. Feller, 'Calcium Carbonate Whites', in A. Roy (ed.), cited in note 17, pp. 203–26. A mixture of calcite and gypsum in the ground on a polychrome statue by Donatello is reported in this article. Two tentative explanations are put forward: the gypsum may have been overheated forming CaO which eventually converts to calcium carbonate; alternatively, the calcite may be an impurity in the mineral gypsum.
67. Calcite identified by X-ray diffraction, in agreement with JCPDS file no. 5–586.
68. Rearick, cited in note 3, p. 102. A fresco of the Venetian Triumph in the Guise of Neptune by Veronese for the façade of the Palazzo Erizzo (now lost) is documented by a drawing now in the Louvre.
69. Gettens, West Fitzhugh and Feller, cited in note 66. It is unlikely that raw limestone was used; the calcium carbonate may have been in the form of lime plaster (slaked quicklime), which over a period of time forms calcium carbonate by reaction with carbon dioxide in the air. See Gettens et al. for a description of the preparation of lime for ‘buon fresco’.
71. Drawings for the figures are reproduced in Rearick, cited in note 3.
72. Malachite identified by X-ray diffraction, in agreement with JCPDS file no. 10–399.
73. Penny and Spring, cited in note 17.
74. Plesters, Roy and Bomford, cited in note 14.
75. Lead-tin yellow 'type II' identified by X-ray diffraction, in agreement with JCPDS file no. 17–607.
76. See Corbeil and Helwig, cited in note 35. Pararealgar was identified by X-ray diffraction, in agreement with the data in this article (JCPDS file no. 33–127).
77. HPLC analysis by Jo Kirby, see p. 71 in this ‘Bulletin’.
78. See Jo Kirby, 'The Identification of Red Lake Pigment Dyestuffs', p. 63 and p. 71 in this ‘Bulletin’.
79. L. Lazzarini, 'I materiali e la technica del ‘Convito in casa di Levi’ di Paolo Veronese', ‘Quaderni della Soprintendenza ai beni artistici e storici di Venezia, Il Restauro del Convito in casa di Levi di Paolo Veronese’, Venice 1984, pp. 65–72.
80. Pignatti, cited in note 36, I, no. A. 62, p. 117; II, fig. 777. Pignatti is not unusual in questioning the autograph status of this work.
81. Ibid., I, no. 343, p. 168; Rearick, cited in note 3, no. 103, pp. 198–200.
82. Inv. 447. See the entry by G. B. Molli in ‘Da Bellini a Tintoretto. Dipinti dei Musei civici di Padova’, ed. A. Ballarin and D. Banzato, Padua 1991, no. 118, pp. 198–9.
83. The earliest surviving paintings on slate are portraits by Sebastiano del Piombo of Clement VII (J. Paul Getty Museum, Malibu) and of Baccio Valori (Palazzo Pitti, Florence), and a battle painting (private collection, unpublished) by Girolamo da Treviso. For a general survey of paintings on stone see the introductory essay by Marco Chiarini in the catalogue of the exhibition ‘Pittura su Pietra’ at Palazzo Pitti, May – June 1970. In Venice, Titian painted an ‘Addolorata’ on slate (Prado, Madrid), sending this to Charles V in 1554, doubtless in emulation of Sebastiano's work, which was known to the Spanish Court. The large altarpiece by Federico Zuccaro dated 1564 in the Grimani Chapel of S. Francesco della Vigna is painted on stone (said to be marble) and later altarpieces executed in Rome by Federico Zuccaro and others are on sheets of slate joined horizontally. Slate was also established as a popular support in Florence by the 1560s. Sebastiano is clearly stated by Vasari to have used coloured marbles and porphyry as well as slate supports, and these may well have been left partly exposed. Such was certainly the case in many small paintings of the late sixteenth century. One of the earliest among these is an oval ‘Adoration of the Magi’, painted on lapis lazuli which is left bare for the blue sky, which has been convincingly attributed to Jacopo Bassano and dated to the 1560s by G. M. Pilo (‘Arte Veneta’, XXIX, 1975, pp. 167–73); and there is also a pair of octagonal. ‘Adorations’ (one of the Kings, the other of the Shepherds) by Jacopo Bassano, of similar date (with Piero Corsini in New York in 1986), painted on an orange pink ‘breccia’ (stated in the dealers' catalogue to be Verona marble) where this serves as the sunset sky. A painting on black stone by Jacopo Bassano is mentioned in an inventory, and that he painted on this support is referred to by early biographers. Given the example of these paintings on lapis lazuli and ‘breccia’ it would seem likely that he experimented with leaving the black stone to stand for the night sky. However, we know of no surviving painting earlier than Veronese's ‘Crucifixion’ in which this device is found – it became, of course, a standard feature in works by later Veronese artists in this mode – Alessandro Turchi, Felice Brusasorchi, Pasquale Ottino, Marcantonio Bassetti, Giambattista Rovedato – whose work is well represented in the Castelvecchio Museum, Verona, as well as in the paintings by Bramer, Stella and others made elsewhere in Italy.
84. Pignatti, cited in note 36, I, no. 341, p. 167; II, figs. 715–16; Pallucchini, cited in note 36, no. 230, pp. 152, 186; Rearick, cited in note 3, no. 91, pp. 178–9.
85. Rearick, cited in note 3, no. 74, pp. 143–4 for a just estimate of this marvellous painting, much underestimated by earlier scholars. A dark grey layer can be clearly seen in the cracks (e.g. in the white dress and flesh of the Princess). The great ‘Calvary’ of the Louvre is also painted on a grey preparation but one which is probably less dark.
86. H. Brigstocke, ‘Italian and Spanish Paintings in the National Gallery of Scotland’, 2nd ed., Edinburgh 1993, pp. 196–7, no. 339, and fig. 62 (X-radiograph).
All rights reserved. No part of this publication may be transmitted in any form or by any means, electronic, or mechanical, including photocopy, recording, or any information storage and retrieval system, without the prior permission in writing from the publisher.
First published in Great Britain in 1996 by National Gallery Publications Limited
5/6 Pall Mall East, London SW1Y 5BA
British Library Cataloguing-in-Publication Data
A catalogue record for this journal is available from the British Library