Paradiso is saturated with stunning light imagery and provocative investigations of visual perception. The enigma of the moon spots and the mirror experiment cunningly provoke the reader into thinking in scientific terms, but reading metaphorically--even codicologically--may yield additional results. The raro/denso, grasso/magro descriptions may refer to parchment which, in turn, brings to mind images of book production. Dante's concern for the integrity of his poem demanded that individual copies be bound. Knowledge of paper manufacturing may help solve the feltro enigma (Inf. I, 105): felt was used to dry freshly molded sheets.
Critical interpretations have considered parchment, paper, scribal transmission, and binding, but one element of book production has been overlooked. To create his ethereal celestial realm, Dante may have drawn inspiration from the remarkable technology pioneered by Italian manufacturers in the decades preceding the composition of his Commedia: the watermarks found in high quality paper.
Although Arabian paper had been produced in Spain and was used in Sicily by 1109, its thick, poor quality sheets were prone to disintegrating. Various improvements attributed to Italians permitted the production of thinner paper with more compact fibers that was translucent when held to the light. Making wire by pulling or extruding it instead of hammering was another European technological advance--datable to c. 1100--that would facilitate the creation of watermarks. Early designs were rudimentary, but they demonstrate a noticeable increase in complexity throughout the fourteenth century as pulp was mechanically beaten and drawn wire became still finer. Such marks were a pragmatic way for the manufacturers to identify paper stocks for sale and shipment. Recalling Dante's raro/denso passage, defects in hides produced thinner spots in parchment; in paper, the thinner areas were man-made watermarks.
Paper would not have been that rare in Italy by the time Dante began writing the Commedia. Among the earliest documented uses of paper is a collection of notarial protocols by Giovanni Scriba in Genoa dated 1154-65. Notaries used paper mostly for drafts and registers, not for official instrumenta publica. Documents from 1264 attest to paper originating in Fabriano, but paper was produced in other Italian cities as well: evidence shows papermaking in Genoa in 1235, and in Bologna papermakers' guild statutes survive from 1255. The idea of copying entire literary works onto paper was still new, but this does not preclude Dante's familiarity with paper and the watermarks used to identify its manufacture.
To study the correspondence between Dante's imagery and watermarks, I first listed the nouns used in Paradiso, eliminating references to abstractions (virtù, ragione), illumination (luce, ombra), sound (armonia, tuono), and movement (danzando, girare) that would be impossible to represent with wire. I discarded the non-visual meanings of words which elsewhere have concrete forms--corno may mean a curved shape or a hunting horn--and excluded nouns never used as watermark symbols: buildings, foodstuffs, gemstones, meteorological phenomena, physical textures, human figures, and topography (one exception: the ubiquitous three mountains). Hunter's watermark classification omits Roman alphabet letters (counted as one item in my Appendix). By consolidating variant spellings and synonyms, this method of analyzing the lexical data produces a total of 122 items capable of visual representation. Next I compared these nouns to published repertoires. Of the 122 images in Paradiso that also belong to watermark categories, a remarkable ninety-one (75%) were watermarks. Of these, forty-two date to 1316 or earlier, the year by which Dante had probably begun composing the Paradiso; six others date 1317-21.
Regarding the reliability of watermarks for dating, this is still an imprecise science and no complete repertoire exists for the thousands of marks once used. Tracings made from dated archival documents provide at best a terminus ad quem for a mark's use. A single design could be used for decades: one eagle was used for 469 years! Thus a watermark from a post-1321 document may actually have been in use during Dante's lifetime and awaits discovery.
Previous studies of the pictorial forms in Paradiso indicated scattered sources without offering a broadly conclusive theory. Luigi Malagoli links the marked pictorial character of Paradiso to medieval spirituality and to the choreographic element of drama. In discussing the visually rich Canto XVIII, Lucia Battaglia Ricci observes: "Aquile, alberi, croci ecc. riempivano i capitelli e le facciate delle chiese romaniche." She pinpoints a mosaic at Ravenna as the source for the cross in the sphere of Mars and notes that the eagle was common in sacred sculpture and heraldry. Yet she offers no compelling reason why these particular images were chosen or why they appear in such close proximity. Judson Boyce Allen suggests that the "stack of icons" Dante encounters during his ascent--the cross, Diligite iustitiam..., M, eagle, ladder, and rose surmounted by the interlocked circles that contain the image of man--reflects a new-style Eucharistic chalice invented in thirteenth-century Siena. Birds, beasts, and angels did appear in the medallions on the bases of such chalices. However, this intriguing structural argument does not accord well with the emphasis on light and illumination in the third canticle. The bright enamelled colors--typically featuring blues and greens--that decorated chalices clash with Dante's prevalent heavenly colors: white, gold, and red.
I wish to offer here a more comprehensive source that accounts for the greatest possible number of images, bearing in mind that the celestial portion of the poem demands visual representations executed in ethereal media (light, motion, and music). Most of the prominent images in Paradiso were very early and common watermarks. Furthermore, of the dazzling parade of images (XXIII, 73-100) that culminate in Peter's keys (XXIII, 139) and the benedetto Agnello (XXIV, 2), all but two--facella and lira--are identifiable early watermarks. Certainly, many of these images were part of an older iconographical tradition of sacred and decorative arts that the craftsmen who formed the watermarks inherited. It is not only these images that would have attracted Dante, but their luminous appearance on paper, "activated" as it were by the sun's rays, whether striking the page from above or behind.
 For an Aristotelian explanation, see Bruno Nardi, "Il canto delle macchie lunari (II Par.)," L'Alighieri 26 (1985): 21-32.
 Richard L. Hills, "Early Italian Papermaking, A Crucial Technical Revolution," in Produzione e commercio della carta e del libro secc. XIII-XVIII: Atti della "Ventitreesima Settimana di Studi," 15-20 aprile 1991, ed. Simonetta Cavaciocchi, Istituto Internazionale di Storia economica "F. Datini," Prato, ser. II, vol. 23 (Firenze: Le Monnier, 1992), 76.
 Ibid., 89. For photographic examples of wire watermarks attached to molds, see Dard Hunter, Papermaking. The History and Technique of an Ancient Craft, 2d ed., rev. (New York: Knopf, 1947), 117, 121, 271, and passim.
 Giancarlo Castagnari and Nora Lipparoni, "Arte e commercio della carta bambagina nei libri dei mercanti fabrianesi tra XIV e XV secolo," Atti e memorie della Deputazione di Storia Patria per le Marche 87 (1989): 188; Hills, "Early Italian Papermaking," 75.
 At least one codex of the Commedia was copied on paper during Dante's lifetime. See Marcella Roddewig, Dante Alighieri. Die göttliche Komödie. Vergleichende bestandsaufnahme der Commedia-Handschriften (Stuttgart: Hierseman, 1984), item 613 dated 1319.
 For recently discovered watermarks, see Le filigrane degli archivi genovesi website: <http://linux.lettere.unige.it/briquet/tipi/>.
 Repertoires and abbreviations: B = Briquet, Les filigranes, op. cit.; G = Archivi genovesi website, op. cit.; L = Luigi Volpicella, Primo contributo alla conoscenza delle filigrane nelle carte antiche di Lucca (Lucca: Dessena, 1911); M = Vladimir A. Mošin and Seid M. Traljic´, Filigranes des XIIIe et XIVe ss., 2 vols. (Zagreb: Académie yougoslave des sciences et des beaux-arts, Institut d'histoire, 1957); S = Francisco de Bofarull y Sans, Animals in Watermarks (Hilversum, Holland: Paper Publications Society, 1959); Z = Aurelio and Augusto Zonghi, Monumenta Chartae Papyraceae Historiam Illustrantia. III. Zonghi's Watermarks (Hilversum: Paper Publications Society, 1953).
 Although not well identified in watermark repertoires, I believe the triangle on a vertical line represents a medieval target. See B unidentified entry #16007: Siena, 1299; and ecu #868: Torcello, 1301.