Een schriftkundig onderzoek van Rembrandt signaturen
R. TER KULLE-HALLER, Hans-Joachim Raupp, W. Frofntjes, H.J.J. Hardy · Oud Holland - Journal for Art of the Low Countries · 1991
Abstract In forensic science, signatures are identified by means of comparative handwriting analysis - not to be confused with graphological examination. To the authors' knowledge, no systematic investigation has hitherto taken place as to the effectiveness of subjecting signatures on old master paintings to such comparative analysis. Even when judgment is passed on signatures in art-historical publications, it is seldom based on an investigation which could stand up to the critical standards of handwriting experts. Partly due to insufficient knowledge of the relevant criteria, signature assessment therefore tends to be influenced in both a positive and a negative sense by opinions about the painting. (I.b.) This article describes the results of examinations of Rembrandt signatures on paintings from the period 1632-1642, conducted by a team of two forensic handwriting experts headed by the controller of the experiment. The Rembrandt Research Project team supplied a total of 123 photographs of signatures, 88 of which were deemed suitable for evaluation, 73 of them belonged to the 'Rembrandt' type, and 15 to the 'RHL van Rijn' type. Only aftcr our examination wcrc we able to confirm, on the basis of Vol umcs 11 and 111 of the Corpus, which of these signatures occur on paintings accepted by the RRP team as authentic Rembrandts, and which on de-attributed paintings. The monograms discussed in Volume I of the Corpus proved to be unsuitable for our investigation, due to the insufficient number of characteristics they yield. (I.d. and I.e.) In the examination of handwriting, the characteristics of a series of incontestably genuine signatures arc compared with the characteristics encountered in a series of signatures whose identity is to be established. This procedure was unfeasible for the examination in question, for, regardless of whether the usual methods of comparison can be applied to signatures rendered with a brush, the question arises as to which of the signatures on a paining are indisputably authentic. Reconnaissance of the signature problem shows that the art historian is unable, for various reasons, to quarantee that a painting established without a shadow of doubt as a Rembrandt actually bears the master's own signature. (I.c.) We therefore opted for a different procedure, but not until exploratory experiments had led us to expect that the usual methods of comparing handwriting would be feasible. (III.a.) Entirely ignorant of art-historical assessments of the paintings in question and their signatures, the hand writing experts analysed the available material on the basis of characteristics used in the comparison of normal signatures and handwriting (11.a.) The team's experiment-controller liaised with the art historians and evaluated the results statistically. (I.d.) The exploratory and statistical character of our investigation was one of the reasons for dispensing with a systematic enumeration of all the individual signature assessments. Working in this fashion, we selected a group of i 'Rembrandt'type signatures from the available material, signatures which formed a homogeneous group because of their shared characteristics. We called them the reference signatures. The homogeneous character of the reference group reflects, in our opinion, the recognizable and reproducible characteristics of Rembrandt's signature. The reference signatures are therefore assumed to have been executed by Rembrandt himself. With the aid of the group of reference signatures, the other material was further evaluated. The outcome was a list in which the signatures are graded as to their probable authenticity. In forensic handwriting comparison, probability gradations stem from the statistical character of the comparison process. (II.b.) They permit nuances to be made in the assessment of a signature. The extent to which identification criteria are satisfied, the consequences of restorations and other doubtful elements which are hard to assess, especially in the case of negative judgment (V.b.), are reflected in the individual probability gradations. Figures 1, 2 and 3 show three signatures from the reference group, accompanied by a number of shared characteristics occurring in the reference group. (IV.a) Figures 4 and 5 show signatures which have been assigned a lower probability gradation; one (figure 4) graded authentic, the other (figure 5) as not. (IV.b.) Only the first four letters of the signature in figure 7 are regarded as authentic; it is one of the seven examined signatures containing only a 't'. (V.b.) This procedure skirts the problem of non-guaranteed comparative signatures. Statistical evaluation of the results can also provide an insight into the question of whether the usual techniques for examining handwriting can be successfully applied to signatures on paintings. If the assessments thus obtained prove to be reliable, they may generate further valuable art-historical information. The results of the examination of the 73 'Rembrandt'-type signa tures are summarized in a table in which the signature assessments are related to the qualifications of the paintings as recorded in Volumes 11 and 111 of the Corpus. This table does not give the probability gradations, which arc however for the sake of convenience simply grouped into 'authentic' and 'non-authentic'. (V.a. and Table I) The table contains the most pertinent statistical data. In order to test thc reliability of the handwriting experts' assessment statistically, we employed a ratio based on signatures occurring on non-authentic paintings. Reliability proved to be almost 90%. Unfortunately, authentic paintings arc not suitable subjects for this kind of test. Evaluation of the results leads us to conclude that, under conditions to be described in greater detail, handwriting examination techniques arc in fact applicable to the assessment of signatures on paintings. The procedure described here only yields results when a large number of signatures with suflicient information content are available. The 73 'Rembrandt'-type signatures permitted the formation of a reference group, but the 15 'RHL van Rijn' specimens were not enough. (V.b.) On the assumption that the handwriting experts' judgment was reliable, only about 40% of the paintings established by the RRP team as authentic were actually signed by Rembrandt himself. It transpired that one of the reference signatures came from a pupil's work (figure 6), as did two others regarded as authentic, albeit with a lower probability gradation. The handwriting experts' results thus supply independent corroboration of the art-historical opinion that Rembrandt signed studio work. (V.b. and Table I) Comparison of the results of our investigation with corresponding assessments of 'Rembrandt'-type signatures by the RRP team yielded 11 controversial opinions: 8 among the 47 authentic paintings and 3 among the 26 unauthentic ones. (V.C.I.) Apart from the experience of the handwriting experts, controversies stem from the fundamentally different approaches of the two disciplines in forming their judgment by means of selecting reference signatures and evaluating characteristics. The fact that the handwriting experts reject more signatures on authentic paintings and accept more in the case of unauthentic ones than the art historians is due to the two disciplines' different kinds of knowledge about the relationship between signature and painting. (V.c.2.) Statistical evaluation of the collaboration of the two fields leads to the general conclusion that the intervention of the handwriting experts results in significantly more rejections of signatures on authentic paintings than has been previously established by the art historians. Moreover, on the basis of results obtained by the two disciplines in the case of the 47 authentic paintings, the statistical expectation is that of all the signed authentic paintings by Rembrandt, roughly half do not (any longer) bear his own signature. (V.c.3) It is to be expected that distinct photographic enlargements, in combination with in situ scientific examination of the material condition of the signature and its direct surroundings, will improve the reliability of signature assessment. Keith Moxey, Peasants, Warriors, and Wives. Popular Imagery in the Keformation, (The University of Chicago Press. Chicago and London, 1989). 165 Seiten mit 57 Schwarz-welß-Abbildungen. In den USA ist in den letzten Jahren eine zunehmende Aufmerksamkeit für die deutsche Kunst der Dürerzeit und speziell der Reichsstadt Nürnberg zu verzeichnen. Die bedeutenden Ausstellungen 'Gothic and Renaissance Art in Nuremberg 1300-1550' (1986) und 'The World in Miniature. Engravings by the German Little Masters' (1988/89) sowie eine Reihe von Dissertationen manifestieren dieses Interesse, dem auch das vorliegende Buch zu verdanken ist. Der Autor hat sich seit seiner Dissertation über Pieter Aertsen und Joachim Beuckelaer (1977) der Erforschung der profanen und populären Bildwclt des 15. und 16. Jahrhunderts in Deutschland und den Niederlanden gewidmet und dabei die Frage nach den gesellschaftlichen Funktionen und Intentionen solcher Darstellungen im Medium der Druckgraphik in den Mittelpunkt gerückt. Der vorliegende Band präsentiert drei Studien zu thematischen Schwerpunkten des Nürnberger Einblattholzschnitts im Zeitalter der Reformation, verbunden durch weitere Kapitel über die historische Situation Nürnbergs und über die medialen Charakteristika von Holzschnitt und Einblattdruck. Der wissenschaftliche Apparat belegt, daß Moxey die reiche deutschsprachige Literatur zur Nürnberger Kunst- und Lokalgeschichte gründlich studiert hat. Seine Übersetzungen der Texte der Einblattdrucke (in den Anmerkungen nach den Originalen transkribiert und zitiert) sind akzeptabel. Das Buch wird mit einer hermeneutischen Standorthestimmung eingeleitet, was angesichts des gegenwärtigen Pluralismus der Kun