The earliest portrait of W.A. Mozart -
A biometrical statistical analysis of the newly discovered Fruhstorfer Mozart from c1762
The painting is in oil on canvas, 49 x 37 cm, apparently still with its first frame, 62 x 50 cm. The portrait was sold in 2010 from a private home about 70 km from Salzburg in Austria, where it had been from 1922. Before that year the portrait was owned by the Austrian brewer family Fruhstorfer. The earlier provenance is not known. It is known, however, that in 1850 Rosina Fruhstorfer and her husband Siegmund Hoffmann became the owners of the pub Bergerbräu in the Linzergasse in Salzburg. This traditional Salzburg pub, which is documented from 1413, was in the neighborhood of W.A. Mozart's home. In 1777 W.A. Mozart was invited to the wedding of the brewer of the Bergerbräu.
1) Photographs of the painting (Figs. 1-4)
2) The status of scientific identification of faces in paintings
3) The three Mozart portraits used for comparison
4) Biometrical statistical analysis
5) Reference images (Figs. 5-10)
6) Where was it painted?
8) The author
9) Update January 2016
1) Photographs of the painting (Figs. 1-4)
Fig. 1 The painting "The Fruhstorfer Mozart" without frame, 49 x 37 cm.
Fig. 2 The painting "The Fruhstorfer Mozart", head and shoulders.
Fig. 3 The painting "The Fruhstorfer Mozart", lower part.
Fig. 4 The painting "The Fruhstorfer Mozart", hand.
Scientific identification of faces in old paintings is still a young discipline. Until a few years ago one could even hear the opinion that the whole discipline was a false beginning. In arts we would be dealing with artifacts, not with images that mirror reality such as in photography. This view, however, is based on a serious category error. From the age of the Renaissance until the beginning of the age of photography, portrait painters worked under a condition of strictly limited freedom. They were commissioned to produce a facial image that would generally be recognized as faithful by anybody who had seen the depicted person. Only in a few exceptional cases could facial traits be arbitrarily altered. As a rule, they had to be reality-bound.
Due to the increasing importance of forensic face identification from
digital photos, and due to the success of scientific face identification
in much discussed cases of Shakespeare portraits and Mozart
portraits, opinions have changed dramatically in recent years. Today
researchers are funded even for the exploration of possible merits of
software-based face identification in arts (1, 2).
The Mozart portrait that would be the historically closest one is the
so-called Lorenzoni Mozart (Fig. 5). It was commissioned
by the father, Leopold Mozart, and is today attributed to the year 1763
and the painter Pietro Antonio Lorenzoni. However, it had to be excluded
as a reference portrait for the present study due to grave painting errors.
In 2006 a newly discovered boy portrait was found to be a partly altered
copy of the Lorenzoni Mozart (3). At that time an investigation
of the case had revealed the following painting errors in the Lorenzoni
A. Landmark test on non-identity
B. Feature test on non-identity
Excursus: Comment on the eye colors of W.A. Mozart
C. Digital feature test on identity
1) Hanging cheeks with a skin fold under the chin.
Excursus: Comment on feature 6
Because the eight features are visible in almost all common light conditions of portrait painting, their frequency in the general population could be determined by feature counts in public portrait galleries.
A corpus of 132 adult Caucasian male portrait paintings was established by extracting naturalistic style portraits that were available in sufficient resolution from the internet archives of the Gemäldegalerie in Berlin, the National Gallery of Arts in Washington D.C., the National Gallery in London, and the Musée du Louvre in Paris.
A corresponding corpus of 108 portrait photographs was established by extracting, in order of listing, the results from Google picture searches that included the search term "portrait".
The count of feature frequencies, separately for paintings and photographs, revealed that in both databases trait 1 and 3 appeared in the same face more often than could be expected from their single frequencies. This observation is biologically plausible, because the probability of both traits is likely to increase with the amount of under-skin tissue in the lower jaw. Therefore, the co-occurrence of these two traits in a face had to be considered as a single new trait.
Similarly, the count of feature frequencies, separately for paintings and photographs, also revealed that in both databases trait 5 and 7 appeared in the same face more often than could be expected from their single frequencies. Also this observation is biologically plausible, because the probability of both traits may depend on qualities of under-skin tissue that are similar above and below the eye. Again, the co-occurrence of these two traits in a face had to be considered as a single new trait.
The feature frequencies were as follows:
1 and 3) Hanging cheeks with a skin fold under the chin AND horizontal wrinkle line between mouth and tip of chin: 8 % in paintings, 6 % in photographs.
2) Vertical indentation in the middle of the lower part of the chin: 46 % in paintings, 31 % in photographs.
4) A nose tip with two tip-defining points: 7 % in paintings, 7 % in photographs.
5 and 7) Dark half-circles (suffused skin tissue) below the eyes AND fold of upper eyelid in parallel to and displaced from edge of eyelid: 8 % in paintings, 9 % in photographs.
6) Left eye more widely opened than right eye: 5 % in paintings, 9 % in photographs.
8) Thinning-out in the lateral third of the right eyebrow: 2 % in paintings, 2 % in photographs.
Next, these six features or feature combinations were tested on correlations. Because all tests were negative and because there is also no biological rationale to assume any correlation, the six features have to be considered as stochastically independent. Thus, the frequency of their joint occurrence is computed by multiplication of the single frequencies. The results for the probability that two non-relatives have the six features in common are one in over 4,800,000 re the painting database, and one in over 4,700,000 re the photo database.
Further, it should be noted that the probability estimate would have
been even stronger, if non-digital features such as the shape of the corners
of the mouth had entered the calculation.
Fig. 5 Five major painting errors in face details of the Lorenzoni Mozart (see sect. 3).
Fig. 6 Global comparisons with the dalla Rosa (1770) and the Edlinger (1790).
Fig. 7 The facial traits used in the present study in the Fruhstorfer Mozart (c1762).
Fig. 8 The facial traits used in the present study in the dalla Rosa Mozart (1770).
Fig. 9 The facial traits used in the present study in the Bologna Mozart (1777).
Fig. 10 The facial traits used in the present study in the Edlinger Mozart (1790).
Inspection of the artist's brushwork revealed a fast but very skillful hand. This indicates a competence and experience that was more likely to be found in Munich or in Vienna than in Salzburg. In 1762 W.A. Mozart was for the first time presented as a miracle boy to the high nobility in Munich and in Vienna. At that time he was six years old, which would be consistent with the apparent age of the boy on the painting. Further, also the likelihood to find the depicted expensive clothing for a little boy and a sponsor for a high-quality portrait was clearly greater in Munich or in Vienna than in Salzburg.
(1) Miller B. Research on Application of Face-recognition Software to Portrait Art Shows Promise. In: UCR Today 2013-05-31, University of California, Riverside. Online publication.
(2) Art research effort aided by face recognition. In: BBC News 2013-06-06. Online publication.
(3) Braun M. Identification of a secondary portrait of W.A. Mozart as a boy: A biometrical statistical comparison with the authentic boy portrait from 1763 attributed to Lorenzoni. In: DN Leeson. The Mozart Cache: The Discovery and Examination of a Previously Unknown Collection of Mozartiana. AuthorHouse, Bloomington, Ind., USA, 2008, pp 216-221. (ISBN 978-1-4343-8415-7)
Martin Braun is a neurobiologist and a composer. He is specialized on
investigating music related auditory physiology. Since 1993 he has published
original research on inner ear function, otoacoustic emissions, pitch
processing in the auditory midbrain, neurophysiology of acoustical sensory
consonance, precognitive absolute pitch, and the physiology of octave
circularity of pitch. From 2000 he works for the independent research
organization Neuroscience of Music near Karlstad in Sweden.
9) Update January 2016
In 2014 the painting was attributed to the Italian painter Gennaro Basile (1722-1782) on multiple stylistic grounds by Barbara Kaiser, director of Schloss Eggenberg and Alte Galerie (Joanneum Graz, Austria). The art historian had special expertise on the portraits by Basile, because she had closely examined a collection of 58 portraits of the nobility of Styria / Austria by this artist (written personal communication, 2014).
A technical analysis of the painting's color pigments in 2014 by Manfred Schreiner, professor and engineer at the Academy of Fine Arts Vienna / Austria confirmed that the same pigments were used as in an established comparison painting by Basile.