Authentication of artefacts and artwork Raman microscopy has been used to 'authenticate' painted objects by identifying the palette used in the creation of the object and noting any date-marker pigments. The latter are synthetic pigments for which the date of first manufacture or availability is well known; their identification therefore provides a terminus post quern for the decoration of the object which can be compared with the purported date of the work. It is important to note that the analysis of the palette of an object can identify materials inconsistent with its suggested date of manufacture, thereby providing evidence of the object's inauthenticity, but can never definitively prove that an object is authentic. An example of such negative evidence of forgery was generated during the Raman analysis of an unusually illuminated thirteenth-century Qur'an for a London auction house; only natural, period pigments were identified . A perfect forgery, although not suspected in this instance, might make use of pigments known in antiquity and other aged materials in order to appear 'authentic,' even if the object were freshly constructed.
Prior to their possible auction, six supposedly thirteenth- to first-century BC illuminated papyri were examined by Raman microscopy and shown conclusively to be fakes . The paintings were found throughout to contain white anatase TiO2, phthalocyanine blue and green, Hansa yellow and Prussian blue; all of these pigments are post-1704 synthetic artists' materials and therefore betray the modern date of the works. A second authentication study by the same authors on a Book of Frames on parchment sought to identify authentic mid-sixteenth-century Ghent/Bruges scatter borders from possible later additions and copies . The collection of eight frames was shown to consist primarily of period mineral pigments with the exception of chrome yellow (PbCrO4), which was only used as a pigment from the beginning of the nineteenth century . However, its presence may indicate extensive restoration after 1846 when the frames were known to have suffered from fire and water damage. This point could only be verified by an art specialist.
Unsurprisingly, researchers have capitalized on the structural specificity of Raman microscopy as a quick, non-destructive means of uncovering faked artefacts and faux jewellery by detecting the use of alternative materials in their fabrication. Raman spectra from purportedly eighteenth- and nineteenth-century scrimshaw (whale ivory carvings) can be used to distinguish those composed of authentic animal ivory from contemporary 'forgeries' composed of substitute materials, such as bone, and from modern fakes made of composite materials [49, 50]. Similar work with precious stones and jewellery has discerned polystyrene, acrylate and polyurethane costume jewellery from true amber, ivory artefacts from simulated ivory, real pearls from faux beads, and true coral from glass imitations [51, 52]. Raman microscopy has been used to verify the presence of genuine ruby and diamond rather than coloured glass on the famous Armada Jewel . Among these precious gems, it is even possible to use Raman spectroscopy to discriminate between true geological specimens and their modern synthetic counterparts, for example those produced using hydrothermal or anhydrous dissolution methods .
Pigments on manuscripts and fine art Most work involving pigment identification on manuscripts and fine artworks has been undertaken as a prelude to restoration or conservation efforts, for authentication purposes (see above) or in the pursuit of art historical information. Because these applications are so numerous and constitute the vast majority of the published research in which Raman microscopy has been used to analyse art, they cannot all be dealt with fully here. Instead, a thematic treatment of the rich information generated by the Raman analysis of artistic palettes will be presented below.
The principal advantages brought to pigment analysis by Raman microscopy are its ease of application, and therefore speed, and its non-destructive nature. The former means that the palettes of large numbers of artworks can be examined in order to test hypotheses regarding the use and history of certain pigments or the techniques and careers of specific artists. The non-destructiveness has allowed pigment analyses to proceed independently of restoration or conservation treatments since the technique is often capable of identifying pigments without the removal of varnishes, either by analyzing serendipitously exposed pigment grains or by collecting the Raman spectrum directly through the varnish . Taken together, these qualities have helped to assuage many of the concerns of curators regarding destructive artefact sampling and long-term removal of an object from display.
The ability to identify pigments with confidence using a Raman microscope requires the availability of a large, reliable database of reference spectra from genuine historical materials. A significant obstacle to the widespread application of the technique in art analysis was overcome with the publication of the first large spectral library of mineral and organic pigments . However, even before its availability, a few investigations had already taken place as the potential of the technique was realized by several researchers working at the Arts-Science interface [7, 55-57]. The first items to be investigated, and still the most frequently examined, were ancient manuscripts [7, 8, 12, 17, 18, 20, 21, 36, 40, 44, 46, 55-74]. This was primarily due to the dominant use of strongly scattering mineral pigments on these artefacts and their watercolour application without an overlayer of varnish. It did not take long, however, before the Raman analysis of all fine art forms had been attempted, including watercolours [31, 53, 75-77], oil paintings [45, 75, 78-80], lithographs [75, 81] and chalk drawings .
The extensive application of Raman microscopy to manuscripts is allowing the compilation of typical regional palettes for Europe [53, 66, 71], Iceland , Scandinavia , Persia , East Asia [8, 64, 68] and the Middle East [40, 46, 67]. Present work by the authors and others will further expand this list and continue to refine these palettes both geographically and temporally. It is expected that these results will add palette identification as an objective criterion to stylistic analysis when attributing problematic manuscript illuminations to specific schools, scriptoria or cultures. One such instance in which preliminary work has been completed concerns the Qazwini manuscripts held by the British Library . In these works, Arabic text is accompanied by a confusing mixture of illuminations in variously Indian, Persian and Arabic styles. Pigment analysis is expected to add to the codicological information regarding the chronology and cultural origin of these illustrations with respect to the script.
Another benefit of defining regional palettes will be the possibility of relating their differences to specific technological, cultural, geological and archaeological circumstances. The anomalous absence of lead-containing pigments, e.g. lead white (2PbCO3.Pb(OH)2) and red lead (Pb3O4), ubiquitous pigments in European illuminations, from a medieval Icelandic work is one such example . Although it is not yet resolved, this peculiarity might indicate inconstant or limited trade links with the European mainland, an observation of potential archaeological importance, rather than simply the artistic prerogative of the Icelandic ateliers.
Pigment analysis using Raman spectroscopy has afforded solid evidence to corroborate and to controvert textual references in treatises that record various illumination techniques. Spectroscopic identification of a thinly applied layer of lapis lazuli (Na8[Al6Si6O24]Sn ) over azurite (2CuCO3.Cu(OH)2) confirms mention of this method of medieval artistic economy
for maximizing the effect of small quantities of the expensive material lazurite, which was difficult to obtain and demanding to refine . Similarly, the medieval concept of hierarchy, and therefore intrinsic value, has been shown to be an important consideration in the artists' choice of certain pigments for the illumination of specific subjects [18, 21, 57]. For instance, in Commentary on Ezekiel, an ecclesiastical manuscript of the eleventh century AD, an abbot and saint are shown together; the abbot's robe is painted in the inexpensive organic dye woad while the revered saint's vestments are coloured richly with lapis lazuli .
The identification of lapis lazuli on this manuscript and chronologically related ones from other abbeys in France is important since it establishes the use of that mineral pigment in Europe nearly two hundred years earlier than had previously been thought [59, 83]. The examination of blue pigments on illuminations from this period shows a transition in which the usage of indigo begins to wane as the dominant blue pigment in manuscript art. Clearly, trade at the turn of the millennium between the East - Afghanistan being the accepted source of ancient lapis lazuli - and Europe was far more extensive than previously believed. The possible earlier availability of this material in northern Europe will be investigated shortly when the blue pigments on the Lindisfarne Gospels from the eighth century AD and other contemporary Anglo-Saxon manuscripts are examined by Raman microscopy through collaboration between University College London and the British Library.
The accuracy of textual evidence for contemporary artistic practices is contestable, and Raman analyses have shown that the information in historical treatises, though of extreme importance, should not be accepted unequivocally. Such a deviation from historical accounts was confirmed by the pigment analysis of the limning in the Armada Jewel given to Elizabeth I by Richard Heneage after the defeat of the Spanish Armada . Although the magnum opus of the foremost limner of the time, Nicholas Hilliard (1547-1619), is quoted as warning against the use of 'ill-smelling colours, all ill-tasting, as orpiment' [53, p. 186], in fact, orpiment was found to have been used on this prized locket.
Although the aforementioned applications of Raman microscopy to the identification of mineral pigments were met with almost immediate success, progress in the detection of organic pigments has been much slower. Organic pigments, dyes and binders on artefacts may suffer from poor Raman scattering efficiency, susceptibility to photo-degradation and intrinsic fluorescence. Only recently has Raman analysis developed to the point of surmounting these obstacles. Significant improvement in the detectability of organic materials has accompanied the recent availability of high throughput spectrometers and more sensitive detectors (see, for example, the improvement in the Raman spectrum of indigo between  and ). Luminescence interference from binders and substrates can be minimized through the use of clever spectral manipulations, such as subtracted shifted Raman spectroscopy (SSRS), which has allowed the detection and characterization of yellow huangbo dye on highly fluorescent papers from the library at Dunhuang, China . Moreover, the increasing use of long wavelength lasers in dispersive and FT-Raman microscopy has further reduced the obstacle imposed by fluorescent species. A wealth of reference spectra for organic materials has been generated using these systems [29, 33-36], and its availability has been lauded as a prelude to the widespread identification of binders, varnishes and organic pigments in artwork using the technique. In some instances, it has even been suggested that the exact species of plant generating the artists' material could be identified from subtle spectral differences .
Despite the enthusiasm, these assertions have been shown by experiment to be overly optimistic for the current state of the art in Raman microscopy as applied to real historical samples. One cause of the discrepancy between the claims made for this type of organic analysis and its successful implementation lies in the reference samples used in the construction of the spectral libraries. In these databases, the reference materials are in most instances single examples of modern specimens that have been analysed in bulk. The organic pigments are represented in the libraries by spectra of the pure chromophore compounds rather than the lakes or dyed substrates likely to be encountered in real samples. Numerous complications therefore arise when one considers that the compositional complexity of the materials, their natural biodiversity, their low concentrations in real artwork, the structural effects of ageing and the effects of ancient methods of preparation are not taken into consideration. These concerns are less significant for mineral compounds since they are rarely affected by such physical and chemical changes from their native state to that encountered in pigments.
Even in the pure specimens used to construct the databases, the clear, distinct differences professed to exist between the Raman spectra of compounds within a specific class of organic materials are often neither clear nor distinct; compare the reference spectra of poppy-seed, walnut and sunflower oils in Vandenabeele et al. . Furthermore, this type of analysis is not as simple as the identification of mineral pigments; the extreme similarity among some spectra of related organic materials requires the use of sophisticated self-deconvolution analysis in order to provide characteristic features suitable for discriminating between different samples . It remains to be seen whether such chemometric analysis will continue to function successfully on real samples. As a consequence of these complications, successful attempts to distinguish between organic binders by Raman spectroscopy are scant, there being only one instance in which the presence of a beeswax coating on paper has been clearly identified . In most instances, it is only possible to provide a general description of the material present, e.g. a 'resin' [74, 85], based on the functional groups revealed in the spectrum.
Although organic analysis has proved to be neither as effortless nor as successful as originally claimed, the ability to distinguish between even broad classes of binder while simultaneously acquiring data on the pigments and dyes contained in an artwork is highly useful and likely to improve with advances in instrumentation and in the quality of spectral databases. As large databases of compositionally similar organic pigments and artists' materials are created, laudable efforts are also being made towards the development of automated search routines for objectively selecting matches between unknown and reference spectra . Various parameterized and non-parameterized modelling algorithms have been tested for the automated deconvolution of pigment mixtures for both qualitative and quantitative analysis of paint layers in artwork; however, the results are not yet convincing .
degradation products are known to be highly sensitive to laser radiation and could transform readily at powers beyond a limiting value. Examples include many of the manganese , iron [27, 28] and lead  oxides, hydroxides and sulfides. This laser sensitivity has led to the mistaken association of a compound with that of its thermally-induced degradation species [40, 67, 73]. In a number of instances, these mistaken identifications have been discovered and corrected afterwards [89, 90]. When examining the important prehistoric black pigment pyrolusite (β-MnO2) however, the laser sensitivity of the material seems to have been unrecognized [74, 91-94]. Various spectra have been presented as genuine β~MnO2, but which deviate from those collected cautiously at low laser power [28, 87].