The Vinland Map --
Some "Finer Points" of the Debate

• Introduction
• The New Tartar Relation
• McNaughton's Disappearing Ink
• Canerio's "Mystery Island"
• Saenger's Diphthong Objection
• Saenger's "Punctilious" Objection
• Saenger's Hyphen-Objection
• Saenger on Shailor's Catalogue Entry
• Saenger's Apocryphal Allusions to other Paleographical Authorities
• Other Saenger Objections
• Technical Considerations: McCrone's Anatase Particles
• Weaver on Anatase in Clay
• Olin's Medieval Ink Preparation
• The Cahill PIXE Analysis
• The Brown and Clark Raman Study
• Enterline's Transfer Mechanism
• Bayes-Cope and Fluorescence
• The Carbon-14 Evidence
• Update 2008
• Seaver's Maps, Myths and Men
• The Partners Bjarni and Leifr
• Seaver on Josef Fischer
• NOVA's "Viking Deception"
• Conclusions and Outstanding Issues
• References

The controversial Vinland Map (VM) is a medieval-style map of the Old World that includes a large island in the western Atlantic identified as "Vinilanda Insula", the Vinland of the Icelandic sagas. It came to light in the 1950s, bound together with a slender and then otherwise unknown tract dating to about 1440, entitled the Tartar Relation (TR). If genuine, the map would demonstrate that mainstream Western Europeans were aware of at least a portion of North America some 50 years before Columbus, and that this information was not restricted to Iceland or even Scandinavia.

In 1957, the map was offered for sale by an unspecified private library through an intermediary dealer to the British Museum, where it was examined by R.A. Skelton, Superintendent of the Map Room, George D. Painter, Assistant Keeper in charge of incunabula (i.e. books published prior to 1501), and Dr. Schofield, Keeper of Manuscripts (Wallis 1974, p. 184). However, the binding was obviously recent, and wormholes in the map did not match those in the TR, even though the subject matter of the TR was reflected in the Asian portion of the map. Because of its lack of provenance and non-original binding, the British Museum turned it down.

Shortly afterwards, Laurence C. Witten II, a Yale alumnus and New Haven book dealer, purchased the volume on behalf of his wife for $3500, and showed it to Thomas E. Marston, Curator of Medieval and Renaissance Literature in the Yale University Library. Marston was at first as skeptical as the British Museum staff had been. However, they soon discovered that a second and much thicker volume that Marston had purchased from the same intermediary, a portion of Vincent of Beauvais' well known Speculum Historiale (SH), must have originally been bound together with both the VM and the TR, with the VM at the front and the TR at the back, since the wormholes in all three aligned perfectly in this configuration, with the minor exceptions noted below. The Speculum and TR both employ an unusual mixture of vellum and paper pages, with two leaves of parchment alternating with every six leaves of paper. The bullshead watermarks on the paper are identical, and the less-than-perfect quality of the parchment is similar. The VM is on a bifolium of rather thin parchment that is essentially identical to two leaves of the Speculum (identified as l1 and l16 by Skelton in VMTR, p. 109).

After this discovery, Skelton and Painter reconsidered their prior doubts about the map, and became enthusiastic supporters. An anonymous donor (later revealed to have been Paul Mellon) paid Mrs. Witten approximately $300,000 for the two volumes and donated them to Yale University's Beinecke Rare Book and Manuscript Library. In 1965, the Yale University Press published, with much fanfare, a volume entitled The Vinland Map and the Tartar Relation (VMTR) with extensive essays supporting the authenticity of the map by Skelton, Marston, and Painter, and an introduction by Alexander O. Vietor, curator of Maps at Yale's Sterling Library.

These authors concluded that the map was made circa 1440, in conjunction with the Council of Basel. The map has an agenda clearly calculated to please the Catholic prelates assembled for the Council: Its captions tell of the Carpini mission carrying the Faith to Tartary in the Northeast, Bishop Eric Gnupsson visiting Vinland in the Northwest, Prester John established in the Southeast, and even Saint Brendan checking out the Antilles in the Southwest.

In 1966, the Smithsonian Institution sponsored a conference to discuss the map. Most, though by no means all, of the sixteen authors presenting papers at this conference were inclined toward authenticity. Einar Haugen of the Harvard Department of Germanic Languages and Literature concluded, for example, that "unless other evidence is presented that invalidates the map on purely external grounds, it can be authenticated as a reflection of information conveyed to Rome by Eric Gnupsson in the early part of the twelfth century." (PVMC p. 142). Stephan Kuttner of the Yale Institute of Medieval Law called the Vinland caption "a piece of hagiographic composition, done before or in the fifteenth century, for the first bishop of Greenland." (PVMC p. 113). Robert S. Lopez of Yale's Department of Medieval Studies, on the other hand, felt that such an unprovenanced document should always be regarded "as guilty until proved innocent," and that the VM still had not made it over this hurdle. (PVMC p. 31). The Proceedings of the Vinland Map Conference (PVMC) were published in 1971, as edited by Wilcomb E. Washburn, Chairman of the Smithsonian's Department of American Studies.

Opinion turned sharply against the map in 1974, however, after chemists Walter C. McCrone (1916-2002) and Lucy B. McCrone of the McCrone Research Institute announced that their tests revealed high concentrations of anatase titanium dioxide in isolated particles in the map's ink, though not on the parchment (McCrone and McCrone 1974; W. McCrone 1988, 1998). Titanium is a common naturally occuring element, but according to the McCrones, the anatase crystalline form of TiO2 never occurs in nature in the particular size and shape of particle that they found on the map. According to them, such particles occur only in modern pigments, and could not have been manufactured until after 1917. After the McCrone announcement, even the Yale Library conceded that the map was "probably a modern forgery."

Despite the McCrone claim that anatase could not have been manufactured in medieval times, Jacqueline Olin of the Smithsonian's Conservation Analytical Laboratory announced in 1976 that anatase titanium dioxide crystals would in fact be precipitated as a byproduct of the medieval process for making iron oak-gall ink, provided the green vitriol used to produce the ink happened to have been manufactured from the titanium-rich ore ilmenite (Olin, 2000, 2003).

And despite the McCrone claim that such particles never occur in nature, Georgia Tech mineralogist Charles Weaver reported in 1976 that the titanium dioxide that commonly occurs in kaolin clays not only takes on the anatase form, but naturally does have essentially the same particle size and shape as found by the McCrones in the VM ink. Weaver's finding shows that the VM anatase particles could in fact have come from nothing more exotic than clay-rich airborne dust.

Finally, in 1987, a group led by Thomas Cahill of the University of California, Davis, used advanced Particle Induced X-ray Emission (PIXE) techniques to determine that although titanium is indeed often present in the ink, it is to be found only in minute quantities, far less than is consistent with the impression given by the McCrone report. The interdisciplinary Davis group, which had previously tested hundreds of other early manuscripts, had detected comparable titanium in several other undisputedly medieval parchments. Titanium was actually present in detectable trace amounts at five points on the VM parchment itself, consistent with "background" contamination from sources such as flaking modern wall paint or, per Weaver's findings, accidental medieval clay dust. The inked lines themselves typically had about twice this amount of titanium, but this would not be enough for the titanium to be a visible component of the ink. Four of the 33 visible inked lines on the VM that were tested showed no detectable titanium whatsoever, demonstrating that titanium in any form could not have been the base of the ink. Cahill et al. cautiously concluded that "while our work argues strongly against the specific McCrone Associates proof that the Map is fraudulent, we do not claim therefore that the Map is authentic. Such a judgment must be based on all available evidence, cartographic and historical as well as compositional" (1987, p. 832).

Encouraged by these three findings, the Yale University Press declared that the map appeared to be genuine after all, and republished VMTR in 1995, with new favorable articles by Painter, Witten, Cahill with Bruce Kusko, and Washburn (VMTR95). The insured value of the value of the map was reportedly revised to $25 million.

In August of 2002, two new technical articles excited considerable press coverage. One of these articles, by Olin with D.J. Donahue of the University of Arizona and Garman Harbottle of Brookhaven National Laboratories, determined the radiocarbon age of the parchment to be AD 1434 ± 11 years. Although this confirms that the parchment is indeed as old as the SH and TR, it still leaves open the remote possibility that a modern forger found two connected pages of blank parchment in the SH+TR or a similarly dated volume, and drew the map on these.

In the second of these 2002 articles, Katherine L. Brown and Robin J.H. Clark of University College London used Raman microprobe spectroscopy to confirm McCrone's finding of significant anatase titanium dioxide at selected points in the ink.

However, even if there really is significant anatase on the map, contra Cahill, it could conceivably have had a medieval origin, either from natural clay particles per Weaver or from medieval manufacture per Olin. The issue of authenticity therefore continues to turn, as it did before 1974, on the humanistic rather than the technical aspects of the map.

Without attempting to resolve the issue of the map's authenticity, the present note mentions an exciting new discovery by Beauvais scholar Gregory Guzman, and then investigates a few of the points raised in two recent papers by Douglas McNaughton and Paul Saenger. Subsequent sections look in greater detail at some of the technical aspects touched on above. Kirsten Seaver's new book is discussed at length, followed by a brief discussion of the February, 2005 NOVA program on the map.

The New Tartar Relation

The Tartar Relation that had recently been rebound with the VM when it first came to light in 1957 is an account of Friar John de Plano Carpini's 1245-7 mission to the empire of the Mongols. Although this mission is well-known from Carpini's own Ystoria Mongolorum, the Yale TR is a somewhat different version, purportedly by a "Friar C. de Bredia," and completely unknown at the time Yale acquired the VM+TR. George Painter provided a complete transcription, translation, and annotation of the Yale TR in VMTR.

At the 1974 British Library Symposium at which the McCrone results were first made public, Francis Maddison, Curator of the Museum of the History of Science, Oxford, presented what he called "A Sceptical View of the Tartar Relation." Although he confessed to playing to some degree the role of "Devil's Advocate," he stated that "doubts about the Tartar Relation are not solely engendered by the company it keeps." (1974, p. 187) He suggested, for linguistic reasons, that the TR may have been the product, not of a thirteenth century friar, but rather "of a more recent forger who had read widely in Slavonic translation." (p. 191) If the TR is fraudulent, then surely so is the VM.

However, any independent doubts that there may have been about the language of the TR, along with any taint that it may have received from its association with the controversial VM, have now been completely removed by the discovery of a second copy of de Bredia's TR in a library in Luzern, Switzerland, by Beauvais scholar Prof. Gregory Guzman of Bradley University (Guzman 2004). It is bound together with a volume of Beauvais's Speculum Historale, just as the Yale TR must originally have been. The final chapters of the SH draw on Carpini's version of his mission, so that it would have been natural to have bound an alternative account of the mission together with it.

The Luzern TR+SH was copied out in 1338-40, approximately 100 years before the Yale TR+SH, and therefore is the more definitive version of the TR. Although the Luzern TR+SH spent most of its career in a now-defunct Cistercian monastery in Luzern, a legal document reported by Guzman places it in the diocese of Basel in 1420, just two decades before the Yale TR+SH(+VM?) was copied there. It is therefore not improbable that the Yale TR was actually copied from the Luzern TR.

Although the new TR by no means validates the VM, it does eliminate the possibility that its TR-derived captions prove it to be a modern hoax.

However, if the VM is indeed a modern forgery, it is surprising that the forger would have chosen to bind it with what at the time was the unique, and therefore extraordinarily rare and valuable, known copy of a medieval manuscript. It would have been adequate for the forger's purposes to have bound the VM with a commonplace and indisputably genuine manuscript, such as a volume of the SH itself; Marston was able to pick up the Yale SH volume by itself, for example, for only £75. Although the Asian portion of the VM includes details unique to de Bredia's account, these could even more effectively have been taken from Carpini's own record of the mission. Instead he or she chose to reduce the independent value the TR would have had by removing it from its important original context and having it "keep company," as Maddison put it, with the hotly disputed VM.

Guzman is currently working on a detailed comparison of the two TR texts. Unfortunately, he found no map of any kind in the Luzern volume.

McNaughton's Disappearing Ink

One of the "latest words" on the humanistic aspects of the debate is Douglas McNaughton's essay "A World in Transition: Early Cartography of the North Atlantic," in the splendidly produced volume Vikings: The North Atlantic Saga, which was published by the Smithsonian in 2000, in conjunction with its Vikings exhibition program. This prestigious venue makes McNaughton's article appear to be the quasi-official Smithsonian position on the map, despite the actual Smithsonian affiliation of both Washburn and Olin, and Washburn's favorable (though unofficial) essay in VMTR95. The New York Times covered McNaughton's article under the headline "Study Casts Disputed Map as False Link to Vikings" (Wilford 2000).

McNaughton, who is described as a physicist and independent scholar of early maps, raises a number of interesting points concerning the VM. Surely the most remarkable, however, is his revelation that the ink originally on the map when it came to light in 1957 has now almost entirely fallen off:

Even an inexperienced observer easily notes that something is very wrong with this document [the VM]. The reproduction shown from The Vinland Map and the Tartar Relation [first published in 1965] is heavily enhanced to show a map that existed only briefly in the mid-1960s. The actual map today has no black ink lines left, as the black pigment has been falling off at a highly abnormal rate since its purchase in 1957 (Moller 1985). This is at variance with all the other maps we have examined in this chapter that are far older and far more detailed. Ink lines may be lost over centuries, but on the Vinland Map all the ink pigment has fallen away in the few decades since it was revealed. (McNaughton 2000, p. 267)

Figure 2
The Vinland Map in 2003
Beinecke Rare Book and Manuscript Library, Yale University.
Similar photo to Figure 1, but image contrast and saturation not digitally enhanced.
2X enlargement
Megazoomable SID file. Requires free download of ExpressView (MrSID) software from LizardTech.

Figure 3
The Vinland Map in VMTR (1965)
Scan from VMTR courtesy Jim Siebold, Cartographic Images, at
www.henry-davis.com/MAPS/Ren/Ren1/carto.html

The forger of the Vinland Map was probably not aware of this requirement for the ink application. Nor was he apparently aware that the drying oil in the varnish (linseed) would immediately be absorbed by the old vellum, leaving only the resin, chromite, solid drying agents, and TiO2 on the surface. It probably looked very good -- and old -- when completed, but the insoluble elements left on the surface react variously to ultraviolet light and temperature. . . . Within decades of even the most careful conservation practices, the normal oxidation and crystallization of the abietic acid in the resin caused the entire ink structure to crumble and fall away. What is left today is only the smallest particles of the ink that followed the linseed oil into the map's vellum. (Ibid., p. 269)

The British Museum's A.D. Baynes-Cope reported, on the basis of his 1967 examination of the map, that "The drawing is very faded, and the parchment has a 'washed-out' appearance, suggesting that it may have been chemically treated in some way" (1974, p. 209, emphasis added). Birgit Reißland (c. 2002) details a number of such chemical treatments that were used historically in an effort to conserve medieval documents.

G.D. Painter, who was one of the first to see the map when it was offered to the British Museum in 1957, likewise reports that the map appeared to have been literally washed, presumably at the time of its modern rebinding:

The VM ink visibly consists in its present condition of two layers, a yellow-brownish line from ingredients which were absorbable into the parchment, and the remains of a surface layer spontaneously formed by non-absorbable particles of black. A similar characteristic is common and well known in medieval manuscripts and requires no special explanation. In the map this overlying black has been more than 90% removed by five centuries of natural wear and degradation, aided by washing and cleaning at the time of late nineteenth-century repair and rebinding. . . . These procedures would be frowned on today, but at that time were routine titivations. (in VMTR 1995, pp. x, xvi, emphasis added.)

It is true that the lines on the map are nowhere near as black today as they appeared to be in the 1965 photograph reproduced in Figure 3 above. However, Yale Beinecke curator Vietor admitted already at the 1966 Vinland Map conference that that photo "was not very close to the tone of the original. This was done consciously -- a very high contrast film was used in order to bring out the legends and map outlines as best we could for the publication." He apologized for any confusion that may have resulted from this not having been made clear. (PVMC pp. 38-9) Note that Figures 1 and 2 above are made from similar recent photographs, except that Figure 1 has likewise had its contrast and saturation enhanced, albeit digitally.

Whatever the method or motive for the removal of the greater part of the black layer that must originally have been present, it is clear that it was already gone in 1965 and that it has not been falling off at an accelerated rate in the past few decades per McNaughton. Evidently McNaughton was unaware that the 1965 photograph was known (since the 1966 conference, at least) to have been enhanced, and jumped to the conclusion that it represented the map's actual 1965 appearance. In short, McNaughton's disappearing ink theory is a blunder that should serve as a profound embarrassment to the Smithsonian, and in particular to the Vikings volume's editors, William W. Fitzhugh and Elisabeth I. Ward.

To top it off, McNaughton's confident assertion that the VM ink is based on chromite has now been refuted by Brown and Clark, who found no chromite in any of the lines on the map (2002, p. 3659). The McCrones had found some elemental chromium in the ink, but not necessarily the mineral chromite per se.

Although McNaughton (2000, p. 269) has a full page box on "Scientific Studies of the Vinland Map," that cites McCrone's findings at length, he inexplicably makes no mention at all of the contrary Cahill study that led to the renewed endorsement of the Map by the Yale University Press and was featured in VMTR95.

McNaughton's Disappearing Islands

Another, equally ill-advised disclosure by McNaughton is that the patches on the map do not cover worm holes but are really part of a cover-up of its post-fifteenth century origins:

A number of small square patches are found on the map; previously it was thought that these covered wormholes; but the patches in two places are located in areas that match the cartographic location of mythical islands shown on sixteenth-century maps (such as the Canerio) and even seventeenth-century maps. (2000, p. 267)

The second interpretation isn't any better, since all the patches but one come in pairs on the left and right sides, covering matching holes in the parchment. Three of these four holes continue into the front of the Speculum as well as might be expected for wormholes, given that its initial four quires (64 leaves) are missing (Marston in VMTR pp. 4-6). Thus, the hole off Ireland (like its mate in Asia) is oval, but becomes round when it reaches the fifth quire of the Speculum, while the small hole off Greenland (with its mate in the "Great Tartar Sea") does not reach the Speculum at all. The Norway and Azores holes, with their Asian twins, continue into the Speculum with no noticeable change in diameter or shape. (The Azores hole may be seen in Figure 5 below.)

During the presumably late 19th century rebinding of the TR, the original covers of the Speculum + TR volume, which had become detached, were put back on the Speculum by itself, but reversed, with the front cover on the back and the back cover on the front. In 1966, Yale decided to restore the three items to their original arrangement. In the process, modern pastedowns were removed from the covers, revealing, inter alia, that the holes in the map and the back of the Speculum would in fact continue into the covers if they were restored to their original positions (Painter in VMTR 1995, p. xvi). When I inspected the materials, I found that indeed all four of the VM wormhole pairs continue into the original front cover. (Evidently the planned restoration was then aborted, since this front cover is still at the back of the volume.) The oval Ireland hole is present, although it is on a crack in the board that has been glued, and hence is now full of glue. The tiny Greenland hole does not fully penetrate the oak board, but is nevertheless present as an indentation. This juvenile bookworm was clearly a finicky eater! Likewise, the major hole at the back of the Speculum lines up with the major hole through the TR and continues through the original back cover.

Painter writes of the holes over the VM patches, "I have seen plenty of fake wormholes in the course of my professional career. You can see how they have been made with a punch, or charred with a redhot hatpin. These are the real thing, they show the dear little worm's serrated toothmarks and they were gnawed by him after the Map was made, for they visibly go through the writing" (1974, pp. 193-4). So much for McNaughton's island cover-up.

McNaughton would have been well-advised to have actually examined the map in person before making his completely erroneous pronouncement about the function of the patches and the non-existence of the wormholes.

Having thus disposed of the wormholes and their alignment, McNaughton is free to claim that there is no solid reason to associate the map with either the Speculum or the Tartar Relation:

A relationship was suggested by the book dealer who owned The Tartar Relation, and the Yale purchaser, Marston, between the map and the Speculum, but the inks of the map and the Speculum did not match, nor did the subject nor anything except ... the unexceptional coincidence that the pages of both the world map and the Speculum were made of parchment and not paper. (2000, p. 257)

Tartar Relation does not refer to the map at all, yet the supposed proof that the map is fifteenth century is largely based on its being found inside a book of that era. (2000, p. 268)

Canerio's "Mystery Island"

McNaughton is on much more solid ground when he points out a remarkable similarity between the layout of the features in the North Atlantic on the VM, and on the splendid world map of the Genovese cartographer Nicolo Canerio, drawn around 1503 to 1505. The relevant portion of the Canerio map is shown in Figure 4 below, while the full map is online at Jim Siebold's Cartographic Images website.

Figure 4
Detail of Canerio's 1503-1505 World Map
Bibliothèque Nationale, Paris
From McNaughton (2000, p. 264).

According to Siebold, the Canerio map was based, in part, on the rather similar Cantino map of 1502, which reflected the latest, top-secret Portuguese data and had been smuggled out of Portugal by Alberto Cantino for the Duke of Ferrara. The green Atlantic island, which appears on both maps, is known to represent Newfoundland and Labrador. This coast, along with Greenland, had been explored for Portugal by Corte-Real in 1500-1501.

McNaughton argues that this island was, for political reasons, deliberately moved far enough east to lie on the Portuguese side of the Demarcation Line of the Treaty of Tordesillas. On the Cantino map the Demarcation Line is actually drawn in, just west of the island. According to McNaughton,

Whoever drew the Vinland Map took advantage of this Mystery Island theme and on it labeled the mystery island "Vinilanda [or Winilanda] Insula" in keeping with a very popular twentieth-century idea that the Vikings had discovered America.

That Greenland is in the same location on both the Canerio and the Vinland Map indicates that the charting of the latter is based on Renaissance Portuguese cartography. Consequently, the Vinland Map cannot be medieval in origin. The shapes of Greenland and Portuguese Newfoundland have not been slavishly copied onto the Vinland Map anymore than the Bianco Map was slavishly copied onto the Vinland Map, but the locations for Greenland and the island labeled "Vinilanda Insula" were copied. Vinilanda on the Vinland Map is exactly the same distance from Greenland, Africa, Spain, and England as the Mystery Island is from those points on the Canerio Map. Greenland, Great Britain, and the Mystery Island form a distinct triangle on the Canerio Map because of the Treaty of Tordesillas, and this triangle is duplicated on the Vinland Map. (2000, pp. 264-265, McNaughton's brackets)

The Canerio and Cantino maps were among the very first to employ an equirectangular projection, a type of cylindrical projection, in which lines of latitude correspond to horizontal straight lines, and lines of longitude correspond to vertical straight lines. The equirectangular projection is in this respect similar to the Mercator, except that the lines of latitude are spaced equally so that there is a constant vertical scale, corresponding to the horizontal scale along one of the lines of latitude. Figure 1.4 in John P. Snyder's cartographic history Flattening the Earth (1993, p. 7) depicts the globe as it should appear in such a projection.

In the sixteenth century, latitude was easily determined by the height of stars and the midday sun above the horizon, but longitude was much more difficult to measure. With patience, the longitudinal offset of different points on land could be roughly determined a process called lunation. This required measuring the right ascension of the moon relative to reference stars by means of an armillary sphere. Since the moon moves only about 12 degrees relative to the stars in 24 hours, a difference in relative position of 1 degree would correspond to a difference in longitude of about 30 degrees. If angles, and therefore latitude, could be measured to the nearest degree, this method would therefore only determine longitude to within 30 degrees. Over a long period of time, several such measurements could be averaged together for greater precision. More accurate results could occasionally be directly obtained by measuring the local sidereal time of lunar eclipses and planetary occultations, and then comparing these at the different locations. Over the decades and even centuries, several such comparisons could be averaged to improve the measurement even further. However, explorers who did not have years or decades to wait for appropriate celestial events had to rely primarily on dead reckoning to get a coarse fix on longitude. It was not until the 18th century that reliable chronometers were developed that allowed longitude to be measured accurately at sea or while exploring new lands.

I cannot read the longitude scale on McNaughton's photograph of the Canerio map, but I would assume that the longitude of Gibraltar at the western end of the Mediterranean had, after thousands of years of observation by everyone from the Phoenicians on down, been reasonably well determined relative to that of the eastern end of the Mediterranean. Indeed, the Canerio map takes Jerusalem, which is essentially due south of the Lebanese coast, as its prime meridian. Gibraltar is in fact 40 degrees, 14 minutes west of Jerusalem.

Taking this as the map's scale of longitude, the eastern tip of South America is about 7 degrees too far east, not bad for a recent discovery over a long stretch of water. However, the northern coast of South America is about 15 degrees too long, so that Aruba at its northern tip is about 8 degrees too far west. As a consequence, the U.S. eastern seaboard is comparably displaced to the west. Thus, the North American promontory at the approximate latitude of Gibraltar and which I therefore take to be Cape Hatteras is about 10-11 degrees too far to the west.

In Africa, Cape Verde is off by only a degree. However, the Cape of Good Hope (not shown) is 10 degrees too far east, and the Horn of Africa is a full 20 degrees too far east, making the Red Sea almost unrecognizably deformed. The southern tip of India is likewise about 18 degrees too far east.

If we take the southeastern corner of the North Atlantic "Mystery Island" to be St. Johns, Newfoundland, this part of the island should be about 12 degrees further west than it is depicted on the Canerio map. Instead, it has, as McNaughton rightly points out, roughly the same visual location, relative to western Europe, that it does on the Vinland Map. (Note, however, that almost half of the spurious gap between Newfoundland and the Carolinas on the Canerio map is due to the longitudinal displacement of the latter, rather than the former.)

Furthermore, the Canerio map's "Mystery Island" lacks the distinctly northwest-to-southeast orientation of the true Labrador-Newfoundland coast. The northern tip of Labrador in fact lies an additional 12 degrees west of St. Johns. Instead the Canerio island erroneously shows the upright orientation of "Vinland" on the VM.

The worst longitudinal errors in the Atlantic portion of the Canerio map involve Greenland itself. Thus, the map depicts Cape Farewell at its southern tip at 62 degrees west of Jerusalem, or 27 degrees west of Greenwich, when in fact it lies a full 44 degrees west of Greenwich. If it were moved 17 degrees west on the Canerio map to its correct location, it would almost touch the northern end of the "Mystery Island," which in turn should lie another 24 degrees further west than it is depicted. On the precise equirectangular projection of the Canerio map, east-west distances at northern latitudes should be expanded in proportion to the secant of the latitude. Instead, Cape Farewell inappropriately lies roughly as it is depicted, relative to Spain, Africa, and England, on the merely ad hoc layout of the Vinland Map.

Furthermore, Greenland should be much wider than it appears on the Canerio map, because of the same required distortion of east-west distances on its projection. Thus, at the Arctic Circle (the latitude of northernmost Iceland), Greenland spans 23 degrees and hence should appear to be two and a half times as wide as the Iberian Peninsula (9 degrees). Instead, Greenland inappropriately has roughly the same width, relative to features in Europe far to the south, that it does on the Vinland map.

McNaughton is quite correct to argue that the remarkable similarity between the layouts of Europe, Greenland, and Newfoundland-Labrador on the Vinland Map and the Canerio Map must betray a common origin. However, the layout is all wrong in terms of the precise equirectangular system of the Canerio Map, yet not unreasonable in terms of the ad hoc system of the Vinland Map. There is no reason to believe that a modern forger, who was clever enough to have found a parchment that could be positively dated to c. 1440, and to have included on his or her map choice details from an otherwise unknown tract that must have originally been bound with the same parchment, would have been dumb enough to have copied Vinland and Greenland from their grossly erroneous depiction (relative to the intended projection) on the Canerio map, instead of from a modern globe. It would therefore be far easier to conclude that the layout on the Canerio Map derives from that of the Vinland Map, than vice-versa per McNaughton!

In all likelihood, the Portuguese and/or Canerio had access to a map containing the same information about the North Atlantic that 60 years earlier had been spliced onto a Bianco-style world map by the Vinland Map's compiler. Conceivably, this was the VM itself, but more likely it was one of several similar maps, now lost. When Corte-Real rediscovered Greenland and Newfoundland in 1500-01, with a good idea of their latitudes but only a very imprecise notion of their longitudes, the cartographers, who probably had no idea what projection system was used on their source map, must have inappropriately copied these features as they appeared onto the Cantino map, whence they migrated to the Canerio map. If McNaughton is right that Newfoundland on the Canerio map lies precisely where it does on the VM, relative to western Europe, it would not even have been necessary to have fudged it across the Papal Demarcation Line to justify a Portuguese claim.

On the Vinland Map, "Vinilanda" is represented as a three-lobed island cut almost through by by deep inlets. It is generally assumed that these lobes represent "Helluland", a barren land of flat stones believed to have been Baffin Island, "Markland," a forested land identified with Labrador, and "Vinland," a land with grapes believed to have been Newfoundland. (Perkins 1974). The VM is therefore loosely using "Vinilanda" to include all three of the western lands. The lake or bay at the end of the northern inlet would then be Ungava Bay. The upright orientation of the island may just mean that the VM compiler was working from a written source that described them as running from north to south. However, another possibility is that the source map actually turned the island clockwise somewhat in order to keep Helluland from getting inordinately far from Greenland, or Vinland too close to Brittany. Some such distortions of direction are required in all but cylindrical projections of the globe. Note that India (and, I would argue, the east coast of Africa) has similarly been turned a full 90 degrees counterclockwise in the Bianco Map, and almost as far in the VM, in order to improve the fit in Asia. The coast of Norway has similarly been turned clockwise in both maps.

On the Cantino and Canerio maps, the "Mystery Island" terminates at the latitude of Cape Farewell, and therefore also of the northern tip of Labrador. Evidently Corte-Real therefore did actually reach Hudson Strait, but not Baffin Island, before turning back, despite the erroneous lack of northwest-southeast drift to the "island" on the maps. Note that Corte-Real, like the Vikings, was greatly impressed with "Markland's" forestation. He evidently took Belle Isle Strait for one of the numerous deep fjords portrayed on the maps.

One innovation of the Canerio map over the VM is that whereas VM follows the medieval tradition (whether based on conjecture or actual circumnavigation does not concern us here) that Greenland is an island, Canerio incorrectly follows Clavus by trying to link it to Spitsbergen, and possibly even to Norway, at the very top of the map. These latitudes, which probably were indeed impassible in the 15th and 16th centuries, are not charted on the Cantino map.

In short, the Canerio map, far from demonstrating that the VM is a post-fifteenth century forgery per McNaughton, actually provides convincing evidence that the unusual information on the VM was employed by Renaissance cartographers, 450 years before VM's modern rediscovery.

For further discussion of McNaughton's paper, the reader is referred to Guthrie (2000).

Saenger's Diphthong Objection

Another recent discussion of the map, one that Tim Spalding finds particularly persuasive in his excellent website "Vinlanda: The Vinland Map on the Web", is a review article by Paul Saenger that appeared in 1998 in the journal Imago Mundi.

Saenger, the curator of rare books at the Newberry Library in Chicago, raises numerous objections to the 1995 edition of VMTR and its conclusion that the map is authentic. However, he states that "the greatest deficiency of the new Yale volume is its failure to address the palæographical and codicological problems posed since the Vinland map's initial discovery in 1958" (1988, p. 199). Since Saenger's own field is paleography, his views in this area carry particular authority.

Saenger raises three specifically paleographic objections of his own. The first of these is that "the writing out of the æ diphthong for the first declension plural in legend number 63 (p. 138) in the map is both inconsistent with the scribe of the Tartar Relation (who uses only the usual medieval form e), and totally anachronistic in a German gothic 'hybrida' in the mid-fifteenth century. At this date, only an Italian humanistic script would have the æ diphthong written out, and neither the Vinland Map nor the related texts in their script evince any other sign of either humanistic or Italian influence" (p. 200).

As a matter of fact, the æ diphthong had been in use in Gothic scripts since at least 1300 AD, as for example in Kong Valdemars Jordebog (Neilsen 1937, Fig. 21 on p. 97), albeit to write Danish rather than Latin language. This issue is therefore one of Latin orthography, rather than of paleography per se as specified by Saenger.

Figure 5

VM caption employing the æ diphthong --
and one that does not.
Figures 5-11 from VMTR 1995

Although Saenger does not mention it, the VM in fact does not use the æ diphthong consistently, but only for this one caption. Thus, the adjacent but well-established Azores and Canary Islands are identified on it as the "Desiderate insule" (see Figure 5 above) and the "Beate insule fortune," respectively, and caption 53 at the upper right refers to "Terre non satis perscrutate". The VM compiler therefore ordinarily did follow the customary medieval Latin -e convention, contra Saenger, and apparently used the exotic -æ convention only where it was called for by his specific sources.

Contrary to the impression given by Saenger, the VM does in fact exhibit numerous Italian influences in both its language and its cartography, if not in its script per se, and therefore might also in its cartographic orthography. As we shall see in greater detail below, Eric Wahlgren had already in 1966 specifically proposed, on linguistic grounds, "that the unknown [VM] mapmaker was Italian" (PVMC, p. 135).

And on cartographic grounds, map proponent R.A. Skelton and critic Douglas McNaughton insist alike that the VM was heavily influenced by the 1436 planispheric world map of the Venetian (and therefore Italian) cartographer Andrea Bianco, or at least by one very similar to it. In fact, the similarity between the two maps is, according to Skelton (VMTR p. 122), most striking in respect to the shape of precisely these two "islands of St. Brendan." A color photograph of Bianco's Planispheric World Map is available on Marco Capurro's very useful "Altre Mappe Antiche" website

Although Bianco himself refers to these islands (on a separate map in his 1436 atlas) as Santanaxio and Antillia, the adjacent Canaries are called the "Isolæ dictæ fortunatæ sive isole ponentur (?) sancti Brandany," using both the æ diphthong and the medieval -e ending, on the 1367 map of the equally Venetian (and therefore equally Italian) Marco and Francesco Pizzigano, according to Skelton (in VMTR, p. 137; Skelton's question mark). Thus, the VM compiler in all likelihood got the name for his Brendan's Islands, complete with its trendy but not anachronistic æ diphthongs, not to mention their outline itself, directly or indirectly from such an Italian humanist source. It should be remembered that Basel was the center of a major Church Council at the time the TR and presumably the VM were produced. Such a Council would have attracted scholars from all parts of Europe, and in particular from Italy, to Basel, even if the VM compiler himself never made the short journey to Italy. Thomas Goldstein (PVMC, p. 52) in fact specifically proposes a connection between the VM and "the environment of geographic thought of Paolo Toscanelli."

Thus, while the use of the -æ ending in one isolated caption on the Vinland Map, but not in several others, is indeed a curious and noteworthy aspect of the map, it by no means invalidates it as a cartographic product of 15th century Basel.

Saenger's "Punctilious" Objection

Saenger's second specifically paleographic objection to the map is that "The punctuation of the legends in the map, notably the total or almost total absence of the 'virgula', a slanting stroke, to mark inter-sentence pauses (the facsimile is not sufficiently clear to say), is inconsistent with the script of the [TR] text and also anormal for late gothic script" (p. 200).

Figure 6 below illustrates a typical occurrence of such a virgula in the TR, between the words [N]est[o]rianorum and [Q]uibus in Paragraph 5 (MS p. 2, col. 1, 8th line from end). Painter's normalized transcription of the TR (VMTR, p. 59) indeed identifies this point as a sentence break, despite the lack of a capital Q on quibus. The virgula itself is the tall hairline stroke. The faint vertical bar next to it is just bleed-through from a heavy stroke on the opposite side of the parchment.

Figure 6
A virgula sentence divider from the Tartar Relation

Figure 7
A punctus sentence divider from the Tartar Relation

Figure 8
A sentence break from the Tartar Relation with no divider symbol

Figure 9 below shows another example of an expected sentence break in the TR, where the virgula, if present, is so faint as to be absent for all practical purposes. Here the expected break is between [bru-]mali and Inchoatur (TR, p. 1, col. 1, line 2 from end, in para. 2). This example is also from the first, presumably well-drafted page.

Figure 9
A sentence break from the Tartar Relation with divider symbol absent or virtually absent

Figure 10
The unique point on the Vinland Map at which a sentence divider is expected,
between the two words on the second of the three lines shown.
Same resolution as Figures 6-9.

As claimed by Saenger, there is indeed no sign of a sentence divider at the spot illustrated in Figure 10, either in the enlarged photograph in Figure IV of VMTR or when I examined the original VM under 2X and 4X magnification. However, if a divider were present, it would be approximately the size of a gnat's eyebrow on the scale of this caption. To complain of its absence is sheer nit picking. Furthermore, as has been demonstrated above, the TR, whose authenticity is incontestable per Guzman's new finding, is itself inconsistent in its use of sentence dividers, despite the much more generous size of its script. The absence of the divider here therefore does not prove that the VM is post-medieval, nor even that it was necessarily copied by a different scribe than the TR.

Punctilious is defined as characterized by exact conformity to all rules and conventions -- literally that every period (punctus) be in its proper place. I can think of no better word to sum up Saenger's objection to the punctuation on the VM, unless it be, in plain English, picayune.

Saenger's Hyphen-Objection

The third specifically paleographic objection Saenger raises against the VM is, "Also, the 'trait d'union', or hyphen, on the map differs in form from that in the [TR] text" (p. 200). According to Reimer (1998), the only use of the hyphen in the middle ages was to mark words broken at the ends of lines, from the 11th century on. Presumably it had the same general form as the modern hyphen.

Unfortunately, Saenger does not specify where in the VM and the TR the inconsistent hyphens he has in mind are to be found. Naturally, the VM's freely placed legends ordinarily do not require word breaks. In fact, the only place on the VM where Skelton's transcription calls for a hyphen is in the word vo[-]lutati, which is broken between the last two lines of the same minute long inscription whose sentence divider is missing in Figure 10. In the enlarged photograph of this legend in Figure IV of VMTR, no hyphen of any sort is evident.

At the same time, a great many words are broken in the TR, with no hyphen to indicate the break. For example, on page 1, column 1, the word bru[-]mali is split between the third and second lines from the end, with no trace of a hyphen. Perhaps I am missing one, but I can't find a single hyphen in the TR anywhere.

It therefore appears that the allegedly inconsistent use of the hyphen between the VM and the TR, like the missing sentence divider in Figure 10, is yet another non-starter by Saenger.

Saenger on Shailor's Catalogue Entry

Saenger particularly praises Barbara Shailor's "excellent 1985 catalogue of the Beinecke's medieval manuscripts" for its discussion of the Vinland Map and its associated manuscripts. According to Saenger, "A close reading of Shailor's full description of Beinecke MS 350A [the VM and TR] clearly suggests that she, as the Beinecke's cataloguer of medieval codices, had palaeographical doubts as to whether the map was in fact genuine" (p. 200).

I've read Shailor's brief entries for MS 350A and the companion MS 350 (the Speculum) on pp. 183-6 of Vol. 2 of her Catalogue (in fact published in 1987, contra Saenger) over closely several times, and in fact see no indication that she was expressing any paleographical doubts of her own as to whether the map was genuine. She does dutifully note that McCrone Associates had found modern anatase titanium and had concluded that the map was probably a post-1920 forgery. But then she also notes that the Davis group subsequently found only trace amounts of titanium, "a fact that suggests the previous quantitative analysis was not methodologically sound." If anything, the fact that she did not repeat Yale Library's 1974 position that the map was probably a forgery indicates that she was leaving the issue up in the air, at least for the record. She may have expressed doubts about the map elsewhere, but she kept these hypothetical doubts to herself in her Catalogue entries, contrary to Saenger's claim.

Saenger portentously announces that

[Shailor] correctly transcribes the inscription on the verso of the map, which both Marston and Painter mistranscribed and mispunctuated in the original Yale volume. . . .

The mistranslated phrase 'Drawing (delineatio) of the first part, second part, third part of the Speculum' was used in 1965 and again by Witten in 1989 to sustain a purported relationship between the map and the text of the Speculum with which it supposedly had been originally bound. In fact the phrase, as Shailor correctly transcribes and punctuates it, is ungrammatical and therefore incoherent Latin. In my judgment, the internal inconsistency in the mode of abbreviating the root par in pars and either partes or partis (whichever the writer intended) suggests a modern confabulation, likely constructed from tracings of discrete entries for the words pars and partis in Adriano Cappelli's Dizionario di Abbreviature Latine ed Italine. A reasonable explanation of the inscription on the verso of the Vinland Map might be that it was added in modern times, perhaps by a book dealer, to relate the map more closely to the genuine medieval text of the volume that is now MS 350; the hand of this inscription need not necessarily be that of whoever confected the map itself, whose latinity was consistently grammatically correct. . . .

Additionally while delineatio is a known neo-Latin term for maps in the sixteenth century, I have not been able to find it used in this sense in the fifteenth century, the normal medieval expressions being tabula or mappa. (Saenger 1988, p. 200)

Figure 11
The caption on the back of the left page of the VM
Contrast digitally enhanced

However, Saenger misleadingly suggests that this inscription is the sole key to the purported association between the VM and the admittedly 15th century Speculum volume. Although it was the clue that initially suggested such an association, the proof is the continuity of the wormholes between the VM and the front of the Speculum and its original front cover, the continuity of the wormholes between the TR and the back of the Speculum and its original back cover, the identity of the watermarks on the paper in the TR and the Speculum, and the similarity of the parchment used in all three. The highly abbreviated caption itself is just a touch of confirmatory icing on the cake. (Although the VM parchment is thinner that most of that in the TR and Speculum, it is, as noted above, essentially identical to two leaves of the latter.)

Saenger rightly points out that the caption could easily have been added by someone other than the scribe who drew the map, perhaps even in modern times by a book dealer who was hoping a buyer would discover the association between the two volumes. In fact, it appears to be a memorandum from the original binder to himself or his apprentice, comparable to the sequence numbers on the quires of the Speculum. The binder would have had to have known a smattering of Latin, but would not have had to have been as expert as the compiler of the VM or even its copyist. If the memo's grammar is less perfect than that of the map, this in no way detracts from the potential authenticity of the map or even demonstrates that the memorandum is modern.

In any event, it is known that the original volume was taken apart and rebound in modern times, apparently circa 1900 to judge from the calf binding found on the TR and VM. The caption in question could therefore even have been added at that time, in the rebinder's best attempt at Latin, so as to harmonize as much as possible with the material in question. Hence even if the caption is modern, there is no reason to conclude that it is part of a deception.

But just how ungrammatical is the caption? My reading of the caption (using the medieval abbreviation sign ~ that appears after numerals, above words, or as a crossbar on the tail of the p's) is Delineatio 1~ p~s. 2~ p~s. 3~ p~r~ spec~li. In Cappelli's dictionary of medieval Latin and Italian abbreviations, the p~s symbol is indeed a standard 15th century abbreviation for pars (1990 6th edition, p. 291). The p~r~ symbol is, as Saenger notes, almost though not quite identical to a second abbreviation that in fact is p~t~, and was used for partis or partes in the same 15th century (ibid., p. 294). It would seem to me that either partis or partes could equally be abbreviated as p~t~ or p~r~. The latter is not recorded in Cappelli's dictionary, but this obviously was never a complete compilation of every single medieval scribe's creative abbreviations, else how could it now be into its sixth, expanded edition?

Shailor's precise transcription of the caption, not given in its entirety by Saenger, is "Delineacio prima pars secunda pars tertia partis [?] speculi". The question mark in brackets after partis is Shailor's, not mine. Since pars is nominative singular, and partis is genitive singular, this would literally read "Drawing first part, second part, of the third part of the Speculum", if my 9th grade year of Latin still serves me correctly. This indeed is a rather ungrammatical shift in case. Using genitive for all three parts (in the sense of "belonging to") as in the VMTR mistranslation Saenger cites above would be ideal, but nominative singular throughout ("Drawing first part, second part, third part of the Speculum") would be acceptable as a fragmentary bookbinder's memorandum.

However, as Saenger himself points out, p~r~ (or at least p~t~) could equally represent partes, the nominative plural of pars, instead of the genitive singular partis. Shailor herself admitted with her bracketed question mark that she was not certain of partis. With this reading, the caption becomes "Drawing first part, second part, third parts of the Speculum", and the awkward case shift is avoided. It is still true that if the plural is used after "third", the singulars after "first" and "second" become redundant, so that "Drawing first, second, third parts of the Speculum" would have been an improvement, but the caption as written can hardly be called "incoherent" per Saenger.

Saenger darkly suggests that his purported modern book dealer lifted these symbols (albeit imperfectly) from an early edition of Cappelli's dictionary in order to establish a connection between the two volumes. However, a far more reliable way to call a potential buyer's attention to the connection between two volumes being offered for sale at the same time and apparently from the same unspecified library would have been simply to announce in the catalog that both volumes use the same paper, similarly interspersed with parchment, and may therefore have been produced by the same scriptorium, even if the dealer chose, for dramatic effect, to leave it to the buyer to "discover" for himself that the wormholes align. A much easier explanation for the odd abbreviations is that a fifteenth century bookbinder merely lifted them from his repertoire of valid fifteenth century abbreviations (with an improvised but completely natural substitution of "r" for "t" in the fancy plural form), and was merely trying too hard to show off his Latin skills to his apprentices by unnecessarily using both singular and plural symbols.

Ever-punctilious, Saenger claims that Shailor, unlike the VMTR's "amateurish" Marston, correctly punctuated the caption in question, thereby helping to demonstrate its "incoherence" as Latin. It is true that Marston (in VMTR, p. 3) incorrectly puts a punctus after "third part," even though there is no mark there, and a colon after "first part," even though the second dot there appears to be just a stray mark. In fact, however, Shailor gives it no punctuation at all, as may be seen above, despite the perfectly clear appearance of a punctus (which could serve as a modern comma) after "first part" and again after "second part". Furthermore, Witten, in his contribution to the 1995 edition that Saenger is supposedly reviewing, does correctly report the punctuation, for what it's worth, even though he incorrectly reads all three abbreviations as partis: "delineatio primae partis. secundae partis. tertiae partis" (Witten in VMTR 1995, p. xlix). Once again, Saenger doth protest too much.

Saenger, as quoted above, questions the use of delineatio itself in the caption as evidence that it could not have been added before the 16th century. However, if I read the 1879 Lewis and Short Latin dictionary correctly, this word appears already in the 3rd c. A.D. writings of the Christian writer Tertullianus, to mean sketch or delineation, and therefore was fair game in this sense in the 15th century. While I have no argument with Saenger's claim that it was not specifically used to mean map before the 16th century, there is no reason to think that this would be its specific meaning on the reverse of the VM. To the binder, the map was simply some sort of sketch or drawing that he had been instructed to incorporate with the separately written material of the Speculum. To object to the map's identification in this context as a delineatio is yet another red herring on Saenger's part.

Despite Saenger's professed respect for the precision of Shailor's reading of the "delineatio" caption, Shailor in fact gives delineacio, even though Saenger reports the same word as delineatio. From Figure 11 above, it may be seen that it is perhaps a close call whether a t or a c was intended. Certainly -tio would be the classical Latin form. Whichever was meant, however, it is disingenuous of Saenger to claim that Shailor, unlike the VMTR authors, transcribes the inscription correctly and that this demonstrates the latters' paleographic incompetence, when in fact Saenger himself dismisses part of her reading.

One important point concerning the map itself that Shailor does raise in disagreement with the VMTR authors, and that Saenger calls appropriate attention to, is her insistence that although the TR and Speculum were written by the same scribe, the VM text is the work of someone else. A circa 1440 date for the TR and Speculum is not seriously in doubt. Therefore if, as stated by Marston (VMTR p. 8) and accepted by Skelton (VMTR p. 109), the VM and TR were lettered in the same handwriting, then the map is automatically genuine, wormholes or no. But if, as Shailor insists, the handwriting is different, any authenticity the map might have must rest on its other features. A different handwriting would not make it automatically inauthentic, since it would be natural for a map specialist to have copied the map while an ordinary scribe was copying the text, and indeed Shailor makes no such claim. Nevertheless it would place the ball of authenticity back in the proponents' court. (Francis Maddison did raise some issues with the TR's authenticity in his 1974 article, but Guzman's subsequent discovery of an even earlier copy of it, noted above, completely smashes this line of argument.)

Neither Shailor nor Marston gave any specifics to back up their contrary claims, so I regard this important issue as unresolved and as one that should be carefully studied by competent handwriting experts. However, it should be noted that just a few different letter forms would not be conclusive of different penmanship, since there are in fact several inconsistencies within the map itself. Just to take the Magnae Insulae Beati Brandani Branziliae Dictae legend in Figure 5 above, the l in Insulae is different from that in Branziliae, the r in Brandani is different from that in Branziliae, the n in Magnae differs from those in Brandani, and three different forms of the æ diphthong itself are used, even though it would be absurd to think that two or even three scribes penned this one legend. Furthermore, the minuscule d in Brandani differs from the minuscule d in the adjacent Desiderate Insule legend, even though the majuscule Ds in Dictae and Desiderate are identical, despite their inconsistent endings.

If the VM was indeed written by a different scribe than the TR, as Saenger believes, the inconsistency between the occasional -æ endings on the map and the uniform -e endings in the TR is no longer so remarkable as he makes it out to be.

Saenger's Apocryphal Allusions to other Paleographical Authorities

Saenger reports that "An oral tradition, perhaps apocryphal, tells us that the late Bernhard Bishoff on a visit to Yale once saw the map and laughed" (p. 200). Surely it must be Saenger himself who is joking if he believes that gossip about a guffaw is any substitute for a published analysis or even a verified casual opinion of the map. Likewise, "Stephan Kuttner is reported to have been sceptical, but he never published his views" (p. 200). As a matter of fact, Kuttner's views were published, in the proceedings of the 1966 Vinland Map conference. Despite the map's numerous peculiarities, Kuttner concluded, "This points once more in the direction not of forgery but of compiled or conflated readings as the source of our caption writer. A sophisticated modern forger -- if we contemplate for a moment this possibility -- would not be so poor a Latinist as to make Pascali the genitive of Pascalis once he took the trouble of finding the name of the pope who reigned in the years of Bishop Eirik's mission. He would not have produced the impossible scrambling of two ecclesiastical functions.... A modern forger, finally, would not be so clumsy as to designate both Innocent IV (1243-45) and Paschal II (1099-1118) sanctissimus pater noster, that is, as now reigning (or reigning within living memory)." (VMTR p. 112) It of course would not be surprising if Kuttner reversed this position after the McCrone report was released in 1974, as did almost everyone involved in the original debate but George Painter (Painter in VMTR95, p. ix). I'm willing to enter hearsay evidence on his later opinion into the record, but not without knowing who heard it said, what Kuttner's specifics were, and when he took this view relative to the McCrone announcement.

In a veritable tour de force of name-dropping, Saenger objects that "the pre-eminent medieval palæographers (such as Bernhard Bischoff, T. Julian Brown, Jean Vezin, Malcolm Parkes, Paul Oskar Kristeller, Giuseppe Billanovich, Albert Derolez, Richard Rouse and Pieter Obbema) have never formally been invited to voice their opinions on the map as a medieval manuscript, either collectively or individually, although many of these erudites must certainly have passed through or near New Haven" (p. 200). If these individuals indeed had something to offer, they were certainly welcome to invite themselves to see the map, as I myself easily did even without their credentials. The fact that they never bothered to is irrelevant to both the paleography and the authenticity of the map. One could just as unjustifiably conclude that it proves that they were completely satisfied with it. The fact that Yale Library and Yale University Press never invited their opinions reflects perhaps on the thoroughness of the Yale investigation of the matter, but not on the map itself.

Saenger notes as further alleged evidence of VM's paleographical deficiencies that "It should also be remembered that the map was shown to and rejected by the curators of the Department of Manuscripts of the British Library (T. Julian Brown was then Assistant Keeper), before it was offered to Yale" (p. 200). However, Saenger does not mention that Painter and Skelton were also among the experts at the British Library (then a part of the British Museum) who examined the map. They were equally skeptical at that time, and became enthusiastic proponents only after the wormholes in the VM, TR, and Speculum were later shown to align. Saenger gives us no indication of Brown's view after the wormholes came to light. Saenger furthermore does not tell us that one of Painter's own specialties is paleography (Painter 1974, p. 193). The 1957 rejection of the map by the British Museum therefore reflects more on its lack of physical context at the time, than on its paleography as per Saenger.

While we are on the subject of the British Museum's 1957 rejection of the VM + TR volume, Saenger approvingly mentions "John Parker's suggestion that Irving Davis (the English dealer who in 1957 had tried, without success to sell the Vinland Map to the British Library prior to Witten's involvement) might conceivably have planted the manuscript of Vincent of Beauvais in the hands of Witten's New Haven friend Thomas Marston so that Witten could 'discover it' as the codex from which the Vinland Map had been removed." This is an interesting thought, but if true, why didn't Davis make sure that the British Museum staff themselves "discovered" the Speculum connection back in 1957, rather than allowing the map to be rejected there, and thereby devalued to the paltry $3500 his seller received from Witten?

Other Saenger Objections

A further objection Saenger raises to the VM is that "eccentric Latinization of both place and personal names suggests a fake. Particularly egregious is Erissonius for Eiriksson where the normal medieval Latin would be Erici filius or Henrici filius" (p. 199).

In fact, this objection had already been disposed of back in 1966 by linguists Einar Haugen and Eric Wahlgren. Haugen (PVMC, p. 141) points out that the map's use of Irlandia for Ireland, Gronelanda instead of Gronlandia for Greenland, for Vinland and Erissonius in place of the expected Erici for Eiriksson simply indicates that the compiler was a "non-Scandinavian."

Wahlgren (PVMC, pp. 134-5) goes on to propose that the extra internal ("svarabhakti") vowel in Gronelanda, Vinilanda, and Isolanda, plus the conversion of kss to ss in Erissonius itself, not to mention the compiler's lack of comprehension of the patronymic nature of this name, specifically indicates, "that the unknown map-maker was Italian." Saenger's objection to the "eccentric Latinization" of these names thus goes hand in hand with his failure to see the Italian influences in the cartography and orthography themselves, noted above.

Saenger also objects that "the Vinland Map's northern orientation, if indeed it truly reflects a thirteenth century lost original, would be without precedent in the Latin West among non-climatic, pre-fifteenth-century world maps" (p. 199). Although the Vinland and Greenland data on the map would, if brought back by Bishop Eirik, have dated to the 12th century (not the 13th century per Saenger), the Vinland Map itself is (if authentic) as much a production of the 15th century itself as is the 1436 Bianco Planispheric World Map on which it is based. Pre-fifteenth century world maps are therefore already an irrelevant comparison. By 1440, northerly oriented Ptolemaic maps had already been circulating and recopied in Europe for three decades. Indeed, we need look no further for such a map than the very same atlas in which Bianco published his Planisphere. On an adjacent page, Bianco included a north-oriented conic projection of the world based on the Ptolemaic data for comparison to his "modern" map. A copy of Bianco's 1436 Conic World Map may be viewed at http://www.capurromrc.it/mappe/!0119bianco.html. The concept of orienting a map with north at the top would therefore have been well known to any student of Bianco's planisphere. The admittedly novel decision to turn the VM in the Ptolemaic direction was undoubtedly influenced in part by the fact that by the time Vinland and Greenland were tacked onto Bianco's planisphere, it fit the page much better in this orientation.

Saenger does raise several troubling issues concerning the provenance of the map. However, if the map was indeed stolen from some library before it was purchased by Witten, that would simply mean that Yale is not its rightful owner. It would not reflect on its authenticity per se.

McNaughton and Saenger both approvingly cite Kirsten Seaver's work on the Vinland Map, and in particular her 1995 article in The Map Collector. Seaver has expanded on this ideas in her 2004 book Maps, Myths, and Men, which is discussed at length below.

Technical Considerations: McCrone's Anatase Particles

In 1972, McCrone Associates were asked by the Yale Beinecke Library to perform a physical study of the map and its ink (Walter McCrone and Lucy McCrone 1974; W. McCrone 1988, 1998). Their technician, Anna Teetsov, removed 29 small samples from the VM, along with 7 from the TR and 18 from the SH for comparison. One of these samples is shown in Figures 2 and 3 of McCrone (1988), and is approximately 100 X 200 micron (0.1 X 0.2 mm.) in size.

On examination of these samples with polarized light microscopy (PLM), they found that several of them contained isolated "small areas, a few square micrometers" (1988, p. 1011; 1 micrometer = 1 micron = 0.001 mm.) that appeared to be rich in the anatase crystalline form of titanium dioxide. In Figure 3 of McCrone (1988), these appear as black specks, perhaps 5 or 10 micron in dimension.

On training a special micro X-ray diffraction camera with a resolution of about 7 microns on these areas, they were able to confirm the presence of anatase, mixed with calcite and some quartz. Other analyses confirming the presence of titanium and/or anatase were also able to be focused on these tiny particles. Their electron microprobe analyzer was thus able to produce results from an approximately 100 cubic micron (4.5 microns cubed) sample. Their ion microprobe analyzer studied samples ranging from 4 to 10 micron in average diameter.

They were then somehow able to physically isolate some of these tiny particles, typical ones of which are shown in Figures 4 and 5 of McCrone (1988). These are both about 5 X 10 micron in dimension, and under a transmission electron microscope (TEM), are shown to be aggregates composed of small rounded pellets. One of these particles, described as "the smallest possible subnanogram portion" (1 nanogram of anatase would fill a cube about 6 microns on a side) was crushed between glass slides and examined by TEM, as shown in Figure 8 of McCrone (1988). It was found to be composed of rounded and somewhat elongated single crystals ranging in size from 0.03 to 0.5 micron in diameter (McCrone and McCrone 1974, p. 212), and averaging about 0.15 micron in size (McCrone 1988, p. 1011).

According to the McCrones, anatase crystals of such a size and shape are not found in nature, and could not have been manufactured in the Middle Ages, yet they are found in modern titanium-based paints, which were not manufactured before 1917. Larger anatase crystals do occur in nature, but if they were crushed to a powder, they would have neither the uniform size distribution nor the rounded shape found on the VM. They concluded that a forger must have used a modern, anatase-based paint or pigment as a component of his or her ink.

Specifically, the VM ink lines tend to have a fine yellowish border that extends slightly beyond the black line. This is a normal consequence of age in medieval documents, and in fact was originally thought of as an indication of the VM's authenticity. McCrone argues that in order to give the map this appearance of age, the forger actually inked it twice -- first with a titanium-bearing yellowish ink to make the yellow line, and then again with a slightly narrower black line. According to the McCrones, "well over 90 per cent" of the black line was then manually flaked off in order to hide the forger's tracks and to give the map a further appearance of antiquity. They found the anatase particles in the yellow line, but not in the remaining portions of the black line.

Weaver on Anatase in Clay

Despite the McCrone claim that anatase crystals such as those found in the VM ink do not appear in nature, Weaver (1976) reports that the TiO2 that is commonly present in kaolinite clays consists mostly of rounded anatase pellets, ranging from 0.05 to 0.20 microns in size, with 0.10 micron "being a good average value" (p. 216). This is a remarkably close match to the range of 0.03 to 0.5 micron and average value of 0.15 micron found by McCrone on the VM (1988, p. 1011).

Furthermore, Weaver found that while these anatase pellets tend to cluster together in clumps of 2-10 or more, they are often found in larger aggregates composed of tightly packed pellets, ranging from 5 to 10 microns, and occasionally up to 80 microns in diameter (p. 216). Weaver's naturally occuring kaolin anatase aggregates thus correspond well in size with the anatase-rich particles isolated and analyzed by the McCrones.

McCrone and McCrone (1974, Table I) provide detailed elemental analyses of 16 titanium-bearing particles that they found. (These analyses are reproduced, along with 3 additional particles, in Table V of McCrone 1988.) If all the titanium they found were in the form of TiO2, which is 60% titanium, only 2 of the 16 particles could have as much as 50% TiO2, while the average value would be only 20-30%, according to their calculation (p. 214). Thirteen of the 16 particles also contain perceptible aluminum, with a median value of 1-2%, and every particle contains at least as much silicon as aluminum. Silicon by itself could be silica, but aluminum with silicon is consistent with kaolinite. If all the aluminum were in the form of kaolinite, which is 20.9% aluminum, the median kaolinite content of the 16 particles would be 5 to 10%. (Computation corrected 4/1/05.) The McCrones' own elemental analysis of their particles is therefore consistent with the presence of some kaolinite that we would expect to accompany Weaver's anatase aggregates. (In a personal communication, however, Kenneth Towe has pointed out that this Al + Si could equally represent feldspar.)

The preliminary McCrone PLM examination of their ink samples found "low refractive index particles, probably ... minerals like quartz, clays, feldspars, etc." (McCrone 1988, p. 1011, emphasis added.) These particles could as easily have been medieval as modern, and so were not subjected to the detailed analysis received by the anatase particles. However, they show that the McCrones' own analysis was at least consistent with the presence of particles of some form of clay elsewhere in the ink.

Although most common clays do not contain kaolinite as their principal ingredient, Weaver (personal communication 2004) states that he "would not be surprised" if anatase pellets similar to those he found in kaolin existed in other clays formed by weathering.

Although anatase-bearing clay dust could easily have blown in an open window and accidentally settled on the parchment, it may actually have been introduced into the writing process deliberately. According to Pete Goebel of Goosebay Workshops, a calligraphic supply e-store, "flour of clay, preferably light in color" (as is kaolin), was one of some 20 substances known to have been used as "pounce" to blot wet ink and/or to prepare the surface of parchment before writing (personal communication). Kaolin clay is particularly prized as an absorbent. Garman Harbottle has cogently suggested that any clay particles that came into contact with the water-based ink would have become suspended in it, and then have become fixed to the parchment as the ink dried. However, any such clay particles on the dry parchment would have been free to fall off as the parchment was handled, whence the absence or relative absence of detected anatase on the parchment itself (personal communication).

Furthermore, Weaver (personal communication 2004) notes that today kaolin is often added directly to ink, as a pigment in its own right and in order to give the ink body. Although none of the medieval iron ink recipes I have seen actually calls for the addition of clay, this simple amendment could easily have been employed.

In 1974, the McCrones could not have known of Weaver's 1976 finding of VM-like anatase particles in kaolin clays. However, neither McCrone (1988, 1998) nor Towe (1990) so much as mentions Weaver's article, despite its great potential relevance to the presence of anatase in the VM. In 1988 or 1990, the relevance of Weaver's finding could perhaps have been overlooked by McCrone, and perhaps even by Towe despite the fact that he is a specialist on clays. However, private researcher Ardell Abrahamson called Wilcomb Washburn's attention to the Weaver study, who in turn cited it in his contribution to the 1995 edition of VMTR (p. xxvi). There is therefore no excuse for McCrone's continued insistence in his 1998 paper that the VM anatase particles could not have had a natural origin.

In private communications, Kenneth Towe has objected that clays contain at most 3% TiO2. Weaver used a calgon dispersing agent, intense sonic treatments, and a centrifuge to separate anatase TiO2 from the bulk of the clay. These methods were not available in the middle ages, yet McCrone found anatase clusters on the VM that were not at least 97% kaolinite. Unfortunately, Towe did not take the opportunity to make these points, as even a brief aside, in either his 1990 article, or in his 2004 reply to Olin's 2003 presentation of her ink argument.

However, if clay were broken up as air-borne dust or pulverized into "flour" with a mortar and pestle, or if it were suspended in aqueous ink, Weaver's anatase aggregates could easily have become separated from the bulk of the kaolinite overburden. In any event, the clusters analyzed by McCrone Associates were isolated and even mechanically separated from whatever background material was present by heroic 20th century laboratory methods, and even then were only 20-30% anatase. This isolation and mechanical separation would not be a commercially viable way to produce even highly impure anatase pigment, but it was performed in a modern laboratory, not in medieval times. As noted above, McCrone Associates did find some aluminum and silicon in most of these clusters, which would be consistent with an origin from clay. They even noted the presence of separate particles consistent with clay, but did not deem these worthy of further study at the time.

It should be noted that Weaver himself has indicated that he would expect anatase from clay to be associated with more iron than typically was reported in the McCrone analyses, since a strong magnet is ordinarily able to remove most of the anatase from the surrounding clay (personal communication, 2004). Nevertheless, in his 1976 article (p. 216 and Fig. 12) Weaver indicates that some of the TiO2 aggregates are not removed by the magnet. These aggregates, at least, would contain relatively little iron.

In further communications, Towe has provided me with a 1974 letter from him to Alexander O. Vietor (Towe 1974, linked below) in which he mentions that Karl Turekian had already raised the possibility that the VM anatase might have come from weathered soils. In this letter, Towe did raise the objection to this possiblity that clay minerals would have shown up in the X-Ray diffractions, yet did not. Although he never mentioned this aspect of the problem in his journal articles on the map, he did refer to "the presence of a well-crystallized anatase free of clay mineral contaminants" in a 1996 letter to the Washington Post (Towe 1996a), and made similar remarks in a letter to the Washington Times (Towe 1996b), both linked below. [Paragraph added 8/11/06.]

Olin's Medieval Ink Preparation

Despite the McCrone claim that anatase particles such as those found in the VM ink could not have been manufactured in the middle ages, Olin (2000, 2003) finds that if the standard medieval process for making the ink ingredient green vitriol (aka copperas, aka ferrous sulphate) is applied to the mineral ilmenite (FeTiO3), anatase titanium dioxide precipitate is one of the end products. This anatase would then turn up in the ink. Her findings were first reported at a 1976 conference, and have been discussed extensively since, even though they were only first published in 2000.

W. McCrone (1988, p. 1017) acknowledges that Olin's reaction produces very finely divided anatase, but dismisses its relevance on the grounds that "to convert this product to a VM-type anatase requires a calcining step at very high temperatures (800-1000 deg. C). This is inconceivable as a 15th century process." Cp. also Towe (1990, pp. 84-85); Brown and Clark (2002, p. 3660).

George Painter, however, has pointed out that "in the real world such metallurgical temperatures were not only familiar to medieval chemists and alchemists but have also been evident ever since the prehistoric bronze-smiths of 3000 B.C., even in the Norse-American context of the L'Anse-aux-Meadows charcoal-fueled smithy, already cited for its titanium-trace slags, where a furnace temperature of 1200 deg. C. has been estimated" (in VMTR 1995, p. xii). Recall that calcination per se is simply the age-old process of roasting limestone to make lime, requiring temperatures of 1000 deg. C. according to Heinemeier and Jungner (1994, p. 35). "Calcinacioun" is in fact specifically mentioned c. 1387 in Chaucer's Canterbury Tales ("The Canon Yeoman's Tale," line 251). Wallace (2000, p. 212) confirms a 1250 deg. C. temperature from L'Anse aux Meadows.

As it happens, medieval ink recipes commonly call for reducing the green vitriol to a powder before mixing it with the other ingredients, in order to facilitate dissolution, and at least one such recipe actually calls for the green vitriol to be "calcined to whiteness" and then "triturated" (ground) for this purpose (Carvalho 1904, Ch. 21). Nevertheless, Kenneth M. Towe of the Smithsonian's Department of Paleobiology (personal communication, Jan. 2004) has pointed out that while temperatures below 300 deg. C. would be adequate to reduce green vitriol to a powder per Carvalho's recipe, at only moderately higher temperatures and far short of those required to convert the anatase particle size, the vitriol would break down into dark red polishing rouge, known in the medieval ages as colcothar, and would become useless for making ink.

Therefore, although medieval technologies could have accidentally produced anatase particles of the size found on the map, despite McCrone's assertion to the contrary, there is no compelling reason to think that these would have ended up in the ink.

Ardell Abrahamson has recently called to my attention an article in which J.C. Deelman (1979) reports that well-crystallized anatase can be produced from amorphous TiO2 at room temperature (25 deg. C), through a repeated cycle of wet and dry conditions, so that the calcining temperatures demanded by McCrone would not in fact be necessary to enlarge the particle size in Olin's precipitate to that found on the map. However, since some 1500 hourly wet-dry cycles were necessary to achieve these results, it seems unlikely that green vitriol made from ilmenite would have been subjected to this process prior to preparing ink. It is possible, however, that something like Deelman's mechanism accounts for the natural presence of well-crystallized anatase particles in kaolinite clays.

Olin (2003) goes on to stress that the elements copper, aluminum, zinc and gold that are present in the VM ink strongly suggest impurities that would have arisen during genuine medieval ink production methods, but not in a modern replica of medieval ink. She calls for further technical studies of medieval documents to determine the precise composition of their inks.

The Cahill PIXE Analysis

Cahill et al (1987) found at most 10 nanogram/cm2 of Ti in 33 visible ink samples on the map, with a median value of 1.0 ng/cm2. Four of the 33 lines showed no detectable titanium whatsoever (< 0.2 or 0.3 ng/cm2), demonstrating that titanium in any form could not have been the base of the ink. Titanium was actually present in detectable trace amounts at five points on the VM parchment itself (median value 0.5 ng/cm2), consistent with "background" contamination from sources such as flaking wall paint or windborne clay dust particles. The inked lines themselves typically had about twice this amount of titanium (median value 1.0 ng/cm2 for the ink plus parchment combined), but this would not, they argued, be enough for the titanium to be a visible component of the ink.

McCrone countered that "If Cahill's 10 ng/cm2 is concentrated in an ink line, it corresponds to nearly 4 million anatase particles in a 1 cm length of ink line." (1988, p. 1016) McCrone does not provide the details of his calculation, but the following assumptions will yield 2 million particles per cm of ink line, which is reasonably close: 10 nanograms Ti per cm2 of target area per Cahill et al's maximum value; 0.5 cm2 inked line per cm2 of target area; 0.1 cm2 inked line per linear cm of inked line; elongated ellipsoidal anatase crystals with long axis averaging 0.15 micron per McCrone (1988, p. 1011), and both short axes half that or 0.075 micron; TiO2 60% Ti by weight; density of anatase 3.9 gm/cc (per the Kronos, Inc. website). Simply shortening the long axis to 1.2 micron and the short axes to 0.6 micron would in fact yield McCrone's 4 million figure.

Cahill's median Ti value of 1.0 ng/cm2 is far more representative of the Ti content of the ink than his maximal value of 10 ng/cm2. Using it reduces the number of particles to 400,000 per cm of inked line using McCrone's calculation, or 200,000 using mine. Either way, this still sounds like a lot of particles. However, it must be remembered that each crystallite is shorter than the wavelength of visible light (about 0.4 - 0.7 micron), and therefore is unresolvable under even the strongest conceivable visual microscope. If these unresolvable particles were spread out one deep, they would cover less than 0.2% of the inked area, using my assumptions.

However, the anatase crystallites on the map are not spread out as individuals, but rather are clumped together into particles approximately 5 to 10 micron in dimension, as pointed out above. If each of these aggregates is a flake 1 micron deep as suggested by McCrone's Figures 4 and 5, each would contain about 60,000 crystallites under the above assumptions if solidly packed, or perhaps only 15,000, if they average only about 25% TiO2 per the data in Table I of McCrone and McCrone (1974). If there are 200,000 crystallites per linear cm, there would then be only a dozen or so of these mold-spore-sized aggregates per typical linear cm of ink line, hardly enough to be a visible component of the ink. (Most fungal mold spores are 2-20 microns in diameter, and as such are invisible to the naked eye.)

There therefore really is a substantial difference between Cahill's findings and the impression McCrone gives that the anatase is an important, visible component of the yellow line. McCrone's online Vinland Map Update indicates that he was still sticking by his original position shortly before his death in 2002. As he correctly points out, the mere fact that the Cahill group's methods are "high-tech" does not necessarily mean that they are more accurate than his more traditional techniques. Both their studies should be replicated by other laboratories to determine which is in error. Towe (1990) generally supports McCrone, but does not demonstrate where Cahill et al. have gone wrong, if they have.

The Brown and Clark Raman Study

More recently, Brown and Clark (2002), using advanced Raman microprobe spectroscopy, report that they have confirmed McCrone's finding of anatase at at least five points in the yellow line, but not elsewhere on the map, and conclude that a modern origin for the VM is "strongly indicated."

Brown and Clark use an approximately 5 micron diameter laser spot size and do not report whether anatase was found in every target they examined or in just a few out of a great many. The fact that they use this beam to search for other minerals and are able to declare them to be altogether absent suggests that they may have searched the entire map for anatase and found only 5 spots containing detectable anatase. However, in a private communication, Clark has confirmed, without giving details, that their findings are indeed quantitatively inconsistent with the Cahill group's. Nevertheless, given the findings of Weaver and Olin, there is no reason that even an abundance of pigment-sized anatase particles should necessarily cast doubt on the authenticity of the map.

Brown and Clark do search for, and fail to find, traces of ilmenite, which they argue should have been present if ilmenite had been the source of the anatase per Olin's mechanism, given the imprecisions of medieval ink manufacture. Unfortunately, Olin (2003) does not address this specific criticism. It is conceivable that any ilmenite fragments would have been large enough to have settled out at one of the steps in the manufacture, so it would have been useful if Brown and Clark had actually prepared some ink from ilmenite and detected ilmenite in a line made from it. This issue deserves further investigation.

By 2002, Brown and Clark should also have known about Weaver's 1976 article, if only from Washburn's mention of it in the 1995 reissue of VMTR (p. xxvi), and therefore should not have repeated the equally erroneous McCrone claim that anatase crystallites such as those found on the VM do not appear in nature. At the very least, as Raman specialists, they should have known and cited a 1997 paper by Enver Murad of the Bavarian State Geological Bureau, who shows that "Raman spectroscopy allows the detection of anatase in kaolin down to a concentration of 0.02%." Indeed, Murad shows a Raman spectrum for a kaolinite clay which, if added to Brown and Clark's spectrum for the uninked parchment, looks essentially like their spectrum for the inked parchment. Murad points out that the principal Raman line of anatase is so strong that for years researchers erroneously believed that it was a line of kaolinite itself, rather than of the small percentage of TiO2 that commonly accompanies it. Murad's findings suggest that Brown and Clark may be detecting only a very small amount of anatase, even within the miniscule targets analyzed with their microprobe.

Brown and Clark do find with their Raman probe that the VM ink, unlike that of the TR, contains substantial amounts of carbon black. Unfortunately, they do not provide a Raman spectrum for iron gallotannate itself, so that their Raman technology by itself does not tell us whether or not iron gallotannate itself is present. They nevertheless conclude, from the fluorescence behavior of the VM ink, that it is not iron-based.

Brown and Clark finally raise a further argument against authenticity, one that is not directly based on their Raman spectroscopy expertise:

It is generally accepted that prior to the development of the printing press, manuscripts were written in either carbon-based or iron gallotannate inks. A common means of differentiating between them is by the identification of erosion of the parchment or paper substrate by iron ions leeching from the iron gallotannate inks. This causes considerable discoloration and embrittlement of both paper and parchment, often leading to brown or yellow staining and, sometimes, considerable loss of the fabric of the manuscript. This trait is not evident in manuscripts produced with carbon-based inks, which are inherently more stable than iron gallotannates. Had the VM been drawn in a medieval iron gallotannate ink, a yellowing at the borders of the ink such as that seen on the map might have been expected. Knowing that such yellowing is a common feature of medieval manuscripts, a clever forger may seek to simulate this degradation by the inclusion of a yellow line in his rendering of the map. However, this study has shown that (a) the black ink on the VM is carbon-based making the natural occurrence of such a feature impossible and (b) that the VM shows no evidence of embrittlement of the parchment or associated parchment loss.

In fact, however, Eusman himself states that "Aged carbon ink and iron gall ink are sometimes hard to distinguish from one another. Visual examination alone does not provide enough information to identify the ink" (barring actual corrosion, presumably). It therefore appears, contra Brown and Clark, that visible corrosion is not an invariable consequence of iron ink.

According to Walter McCrone, all black medieval ink, not just iron-based ink, tends to develop a yellow stain around the inked lines over time. On p. 30 of his 1998 article, he states, "One thing we know about any black ink lines is that after a few hundred years, a narrow yellow border should be visible along the ink lines due to migration and yellowing of organic components of the ink; these stain the parchment fibers yellow. This is usually a requirement for, and a proof of, age." Thus, whether or not there is corrosion caused by the chemicals in the pigment, the organic components themselves (ie primarily gum arabic in the case of iron ink per Carvalho 1904) cause a yellow stain that gradually spreads.

That this is not just a slip of McCrone's declining years is indicated by similar (though less detailed) statements in his 1988 Analytical Chemistry article: "The black upper line was drawn, and generally well-centered, over a broader yellow line. The latter simulates the natural diffusion and yellowing of an ink medium into the parchment which takes place normally over hundreds of years" (p. 1014), and, "The forger apparently knew that the medium in an old ink line diffuses very slowly over a few hundred years to cause a yellow border along its length" (p. 1015). Again, it is not the iron or even the gallo-tannic acid, but rather the medium (gum arabic dissolved in water that has since evaporated) that causes the stain he has in mind.

This is not to say that the iron pigment itself cannot cause a stain, or even damage, to the parchment. However, unless McCrone is mistaken here, a (noncorrosive) yellow stain would appear with age whenever the same or similar organic medium is used to apply any pigment. Of course, McCrone has been known to be mistaken on other specifics, as for example about the existence of pigment-sized anatase particles in nature, or about the feasibility of calcination during the middle ages. However, until there is a published demonstration that he is wrong, the matter is unsettled.

Eusman states that carbon ink was indeed made with gum, to keep the carbon in suspension, so that the same organic medium was sometimes present in carbon ink as in iron ink. Furthermore, the soot was at most 80% carbon particles, according to him, with the remainder including tarry materials. Tar would impart a yellow stain if it migrated with the gum into the fibers, so that if McCrone is right about the mechanism, a yellow stain could easily form with carbon ink.

On the other hand, Eusman notes that pigments were often added to iron ink, to make it more visible as it was applied. Lampblack would serve this purpose, so that carbon in the ink does not itself prove that the ink was not primarily of the common iron type. Soot could even have become the dominant pigment if in a preservative effort, some chemical was applied that almost completely removed the iron.

It should be noted that according to Carvalho (1904), isinglas (a pure form of gelatin made from sturgeon bladders), rather than gum arabic, was most commonly used to make carbon ink, at least in oriental recipes. Olin (2003, p. 6747) notes that McCrone's infrared spectroscopy of the ink did identify the presence of gelatin, though she notes that this could be "a result of hydrolysis of the collagen of the parchment by the acid present due to the deterioration of an iron gallotannate ink." In summary, the Brown and Clark argument that carbon in the ink, plus a yellow line with no corrosion, by itself demonstrates forgery, has not at present been substantiated.

Clark (2004) critiques Olin's (2003) note in Analytical Chemistry. His principal argument is that the ink is carbon-based, and that therefore Olin's mechanism is irrelevant to the appearance of anatase in the ink. However, he still makes no mention of Weaver's finding of VM-sized anatase particles in kaolinite clay (1976), or of Murad's detection of even trace amounts of anatase in the Raman spectra of such clays (1997). Instead, he expressly states that such particles do not occur in natural weathered mineral samples, in direct contradiction to Weaver.

Enterline's Transfer Mechanism

Yet another mechanism to account for the presence of pigment-sized anatase crystals in the VM ink has been proposed by James Enterline (2002, Appendix).

Enterline observes that the 1937 edition of Plenderleith's manual on conservation methods recommends cleaning manuscripts (presumably for mildew) with sodium hypochlorite, i.e. household bleach. In particular, if the ink is believed to be fragile, Plenderleith recommends first covering the document with tissue paper and brushing the bleach into the document through the tissue paper.

Enterline proposes that if the tissue paper (e.g. old-style typewriter carbon-copy paper) had been whitened with modern anatase pigment and bound with casein, the bleach would soften the casein and some of the pigment could be transferred to the document. In particular, since the bleach would also soften the binder in the ink, which was commonly gum arabic, a type of glue, the anatase particles pressed into the ink would actually become bound into it. On drying these would be retained, while any that adhered to the paper would not.

Enterline reports that when he tried this with an iron-tannin pseudo-ink on paper, using thin paper from a 1952 Bible that would have been bound with casein, he found that anatase pigment particles were retained by the ink, but not by the paper beyond the ink line.

Enterline first proposed his very cogent mechanism at a 1977 cartographic conference, but it was not published until 2002, and has not yet been widely discussed.

Baynes-Cope and Fluorescence

The well-cited anatase debate has drawn attention away from other important technical studies of the map, and in particular from Baynes-Cope's 1974 finding that the Map reacts in an unusual way to UV light:

The behavior of the ink under ultra-violet light was of particular interest. Iron compounds quench the fluorescence induced in the background by ultra-violet light and for this reason, faded iron gallo-tannate ink, yellowish-brown by daylight, will appear black against a bluish or yellowish fluorescent background under this form of lighting. The inks used in both the Tartar Relation and the Speculum Historiale showed this phenomenon whereas the ink used both for the outline of the map itself and for the text on the leaf did not show this phenomenon. (Baynes-Cope 1974, p. 210)

Figure 12 below shows the photo of the VM that accompanied McNaughton's article in the Smithsonian's 2000 Vikings volume. Most of the map's background in McNaughton's photo has a weird blue cast to it, even though the parchment is actually tan. The only exceptions are the unobtrusive patches on the back of the map, which show through brightly as yellow, and a faint smear of yellow in the western Atlantic below "Vinilanda." Evidently the map in this photograph is exhibiting the blue and yellow fluorescence under UV light that was described by Baynes-Cope. The modern parchment patches glow yellow rather than blue like the map, either because they were cured with different chemicals, or, as is more likely in light of the faint yellow smear in the Atlantic, because of the glue that holds them in place. It is therefore safe to conclude that McNaughton's photograph was taken under UV light, even though he makes no mention of this fact.

Figure 12
UV(?) Vinland Map photo in McNaughton (2000)

Several photographs of the VM and TR, including Figure 2 above, may be viewed by navigating to http://beinecke.library.yale.edu/dl_crosscollex/, and searching for Vinland Map. Figure 12 above is clearly based on color photo #39002037848968, while the older B&W photos 39002032765142, ...159, ...407, ...415, ...431, ...449, and ...456 were all taken under the same conditions, since the lines are darker than the parchment, while the patches and slug-shaped blob under Vinland are lighter than the parchment. However, it should be noted that according to Ellen Cordes, Head of Public Services at the Beinecke Library, there is no record that any of these was taken under UV light (personal communication, 2/13/05). The only one of these photos which is actually identified as shot under UV light is the old B&W #39002032765399, in which the patches and "slug," as well as the lines, are distinctly darker than the parchment, unlike Figure 12. It is conceivable that Figure 12 and #...399 are both UV, but using different wavelengths, or that Figure 12 is really infrared. In order to clarify the important issue of how the map really looks under UV, it would therefore be very useful if Yale would add detailed modern color photos like Figure 2 above, taken under UV and IR light of appropriate wavelengths, along with similar photos of say the first page of the TR and SH.

The C-14 Date

As noted above, Donahue, Olin and Harbottle (2002) determined the radiocarbon age of a sliver of the parchment taken from the bottom of the map to be AD 1434 ± 11 years. Although this confirms that the parchment is indeed as old as the SH and TR, it does leave open the possibility that a modern forger found two connected pages of blank parchment in the SH+TR or a similarly dated volume, and drew the map on these.

However, this possibility is rather remote, since the wormholes in the VM parchment strongly indicate that it originally was at the front of the SH+TR volume. Manuscripts, including the Yale TR itself, often do have blank pages at the back end, and books often contain a blank page or two at the front. However, a blank leaf at the front of the first quire would ordinarily be connected to a written leaf at the back of the same quire. The bookbinder might have added a bifolium of blank flyleaves at the front, but ordinarily these would have been in a much coarser material than was used for the manuscript itself.

Since the writing on the map has been actually chewed through by the bookworms, the hypothetical forger would have had to have acquired live bookworms per Lopez (PVMC, p. 31 n. 1), and placed them together with the freshly drawn map in its original configuration between the SH+TR front cover and the SH in order to generate the aligning wormholes. He or she then unaccountably rebound the VM+TR separately from the SH in order to prevent the alignment of the carefully made wormholes from being noticed except through the impulse purchase of the SH on Marston's part, and then even took the trouble to conceal the aligning holes in the original front cover with a paste-down.

It would be far simpler simply to believe that the map is a genuine 15th century production.

Lönnroth (1997) suggests that Vinland and Greenland were added by a modern forger to a genuine 15th century map of the Old World. Harbottle himself, as quoted by Lönnroth, has noted that the ink used for Vinland and Greenland has a distinctly different chemical composition than that used for the Old World. While this is an interesting idea that would account for both the C-14 date and the wormholes (none of which penetrates either Greenland or Vinland), it falls on the fact that the "oval" of the VM Old World has been deliberately positioned off-center in order to accommodate Vinland, Greenland, and their caption. The two western "islands" could then have simply been done with a different batch of ink. In light of Lönnroth's suggestion, however, it would be very desirable to have a forensic handwriting expert compare the handwriting in these two areas, in addition to the TR and SH. (Paragraph added 8/21/06.)

One unexpected finding of the Donahue et al. study was that some 20-30% of the mass of their sample was an organic substance of modern (post-1950) origin. It was only after removing this substance with acetone that they obtained their medieval date on the parchment itself. Apparently this modern substance is a varnish or similar material applied in an attempt at preserving the map sometime shortly before it was first shown the the British Museum.

McCrone, who was not aware that the map had undergone such a treatment, noted that under his initial examination of the map with a stereo microscope, he found that "The black ink line was bordered by a yellowish discoloration which was at first assumed to be the stain normally resulting from discoloration by ink components having migrated into the fibers over time. During sampling, however, this yellow discoloration was observed to have body, unlike a stain. It could be removed as tiny fragments with a fine-tipped tungsten needle while viewing the map with a stereo microscope at 20-60X." (1988, p. 1011) After finding tiny anatase clusters in these fragments, he concluded (p. 1015) that a post-1917 forger painted the yellow line and then added the black line on top of it in order to give the manuscript the appearance of age.

However, this paint-like property of the yellow line that McCrone found to be so suspicious could simply be the effect of the modern organic substance that Donahue et al. found had been applied to the map post-1950. If the anatase particles in the ink lines did just fall off the ceiling as suggested by Cahill, it could account for why they nevertheless seemed to be embedded in the ink, as noted by McCrone technician Anna Teetsov (NOVA 2005). It would therefore be useful to see if similar fragments can be lifted off the map's uninked areas.

Update 2008 (Section added 9/08)

In a new article in Archaeometry, Garman Harbottle (2008a) argues that the calcite crystallites found by McCrone in the map's anatase-bearing particles are inconsistent with 20th century synthetic anatase production, since this involved hot sulphuric acid that would have destroyed any calcite. Furthermore, synthetic anatase pigment commonly contained substantial barium sulphate, which McCrone did not detect on the map. He argues that a natural origin for the anatase, such as kaolinite per Weaver (1976), or even lateritic earth pigments such as umbers, sienas and ochres, is much more likely.

Towe et al. (2