Journal of Archaeological Science 37 (2010) 192–200

Contents lists available at ScienceDirect

Journal of Archaeological Science

journal homepage: http://www.elsevier.com/locate/jas

A dendrochronological investigation of stringed instruments from the collection

of the Cherubini Conservatory in Florence, Italy
Mauro Bernabei a, *, Jarno Bontadi a, Gabriele Rossi Rognoni b
a
CNR/IVALSA, Trees and Timber Institute, National Research Council, via Biasi 75, 38010 S. Michele all’Adige (TN), Italy
b
Dept. of History of the Arts and Performance, Galleria dell’Accademia, University of Florence, Florence, Italy

a r t i c l e i n f o a b s t r a c t

Article history: A total of forty-nine stringed instruments of the Conservatory Cherubini collection, at the Musical
Received 13 January 2009 Instruments Department of the Accademia Gallery in Florence, were submitted to a dendrochronological
Received in revised form investigation in order to date them, check the validity of their attribution and to find out more about
8 September 2009
their construction characteristics. Thirty-seven instruments were successfully dated, thereby deter-
Accepted 10 September 2009
mining the terminus post quem date of manufacture. The correlation values of the statistical cross-dating
tests were generally very high. The dendrochronological analyses determined which instruments had
Keywords:
been made from wood of the same provenance and, in some cases, from the same tree trunk. The mean
Dendrochronology
Violin chronology built from the musical instrument series, named ‘‘Accademia Master Chronology’’, is 558
Stringed instruments years long and dates from 1396 to 1953AD. The interval between the youngest ring dated dendro-
Wood provenance chronologically and the given date of manufacture increased constantly in the course of the centuries,
Violin-makers from a mean value of just over eleven years for instruments built in the eighteenth century, to nearly 74
years in the twentieth century, when the use of old wood from other artefacts became more frequent.
Furthermore, in the Cherubini Collection, the average tree rings on violins are smaller than those of other
stringed instruments; in fact, they increase in proportion to instrument size and are widest in cello and
double bass.
Ó 2009 Elsevier Ltd. All rights reserved.

1. Introduction - The object must be made of a tree species suitable for

Dendrochronology is a scientific wood dating technique that
studies the growth rings of trees in relation to time, which today
represents the principal method of dating and studying wooden
artefacts (Baillie, 1982). There are numerous art-historical appli-
cations of dendrochronology, ranging from panel paintings and
wooden sculptures to antique furniture. In fact, any object made of
wood can be dated, if the following conditions are fulfilled:
- The object must contain an adequate number of clearly visible
and measurable rings. Dendrochronology is based on statistical
correlations: the reliability of the result increases in proportion
to the amount of data (¼the number of tree rings) compared.
Even though there is no universally acknowledged minimum
number of rings for dating purposes (UNI Standard 11141,
2004), generally tree-ring series with a least sixty rings are
considered reliable (Munro, 1984);
* Corresponding author. Tel.: þ39 0461 660207; fax: þ39 0461 650045.
E-mail address: bernabei@ivalsa.cnr.it (M. Bernabei).
dendrochronological dating. Riverside species, such as poplar
(Populus spp.), willow (Salix spp.) and alder (Alnus spp.), or
those subject to severe human disturbance, for example fruit
trees (e. g. Rosaceae sp), chestnut (Castanea sativa Mill.) or
walnut (Juglans regia L.), generally are not suitable for
dendrochronological analysis (Fritts, 1976; Bernabei et al.,
2007);
- There must be a reference chronology for that particular
species and geographical area.
For stringed musical instruments, these conditions are often
fulfilled (Klein, 1985) and, in fact, the earliest dendrochronological
investigations date back to the late 1950s (Lottermoser and Meyer,
1958), although proper dating has been carried out only since the
1980s (Corona, 1980; Klein et al., 1984). Since then, there have been
many important applications of this method (Klein et al., 1986;
Topham and McCormick, 1998, 2000). Mainly stringed instruments,
such as violins, violas and cellos, made by well-known violin-
makers, have been analysed, whose attribution, in some case, had
been questioned (Grissino-Mayer et al., 2004; Grissino-Mayer and
Deweese, 2005).

0305-4403/$ – see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jas.2009.09.031
 M. Bernabei et al. / Journal of Archaeological Science 37 (2010) 192–200
Traditionally, the belly of stringed instrument was made of
spruce (Picea abies Karst.) wood, whereas the back and the ribs were
constructed from maple (Acer platanoides L.) or sycamore (A. pseu-
doplatanus L.), although ancient instruments may also contain less
valuable wood, such as poplar. The fingerboard is usually made of
ebony (Diospyros spp.), whereas the pegs are made of hardwood
such as cherry wood (Prunus sp.). The purfling consists of black-
stained willow, beech (Fagus sylvatica L.) or similar woods. The
bridge is made of maple whilst the linings and the blocks at the
corners may also be in beech wood, willow, coniferous wood and
others. A preference for certain types of wood for specific parts of
the instrument is often a distinguishing characteristic of the various
schools of instrument making: for example, a beech wood purfling
is typical of the Tuscan school. Unfortunately, apart from spruce,
none of the other species is suitable for dendrochronological dating.
The almost exclusive use of spruce wood for instrument bellies
is mainly due to the particular characteristics of the wood from this
species that guarantees, in the best cases, an optimal acoustical
quality of the instrument (Bucur, 2006; Wegst, 2006).
The belly of a stringed instrument is made by radially sawing or
splitting a piece of spruce trunk to produce a wedge, which is
subsequently sawn again with a radial cut, opened like a book, and
joined again along the opposite, bark, side of the wood (Fig. 1). In
this way, the belly of the instrument is made of two symmetrical
pieces, with the youngest part of the wood towards the centre.
There are, however, also cases in which the belly is made of a single
piece of wood or where two pieces are joined the opposite way. Or
else, especially in the case of bigger instruments, such as cellos or
double basses, there may be more than two pieces.
The selection of spruce timber for building musical instruments
follows precise qualitative requirements, such as an absence of
knots, reaction wood or other defects in the wood. Finally, the
violin-maker’s sensitivity and experience play an important role in
the choice of the wooden material.
Hence, the necessity to obtain the best acoustic properties leads to
the use of wood that is free from defects and, thus, incidentally,
presents the best characteristics for dendrochronological research.
Generally, the aim of dendrochronological analysis is to determine
a terminus post quem or the date after which the belly of an instrument
was made, in order to corroborate or refute a potential attribution to
a violin-maker or a school of violin-making. Dendrochronological
examination may also supply other information, regarding the
provenance of the wood or particular idiosyncrasies in the
construction of the instrument (Klein et al., 1986; Corona, 1998).
The purpose of this study, which was set up during the prepa-
ration of the catalogue of the Collection, is to analyse a collection of
great historical interest, like that of the Cherubini Conservatory in
Florence, which consists of an important and homogeneous
nucleus of stringed instruments from the school of instrument
Fig. 1. Schematic representation of the construction of a violin belly following standard procedure.
193
making in central and northern Italy, thereby contributing to place
these artefacts into the correct historical and technical framework.
This work, therefore, has the following objectives:
- To date the stringed instruments by determining their terminus
post quem date;
- To assess the provenance of the wood used by the instrument
makers, on the basis of existing reference chronologies;
- To examine the possibility of creating a reference chronology
from the instruments of the Accademia;
- To provide possible indications regarding the construction
characteristics of the instruments.
2. Materials and methods
2.1. The instruments
The Collection of musical instruments from the Conservatory
‘‘Luigi Cherubini’’ that, since 2001, has been conserved and
exhibited at the Department of Musical Instruments of the Acca-
demia Gallery of Florence, comprises about five hundred instru-
ments, sixty of which belong to the violin family (violins, violas,
controviolini i.e. bass-violins, cellos and double basses).
The original nucleus of the Collection, which is not included in
this study but will be the subject of future research work, consists
of instruments from the private collections of the Grand Dukes of
Tuscany, Medici and Lorena, including, in particular, some speci-
mens still in existence that belonged to Prince Ferdinando de’
Medici (1663–1713), son of Cosimo III and contemporary of Antonio
Stradivari. The Collection holds three instruments made by Stra-
divari, including the only one in the world to have survived
completely intact in all its parts (Antonio Stradivari, tenor viola
‘‘Medicea’’, Cremona 1690; see Falletti et al., 2001).
To this first group, consisting of eleven instruments, another
forty-nine were added (Table 1) between 1863 and 2001, either by
donation or acquisition: twenty-four violins, fourteen violas, three
cellos, two double basses and six controviolini (instruments built at
the beginning of the twentieth century by the violin-maker Val-
entino De Zorzi and playing an octave below the violin, which
places them between viola and cello, both by tuning and size).
Although German and French violin-makers are also repre-
sented, the Collection mainly consists of Italian instruments made
between the second half of the eighteenth and the first half of the
twentieth centuries.
The nucleus attributed to Florentine and Tuscan violin-makers,
of which this Collection includes the largest group presently
known, stands out in particular (Rossi Rognoni, 2004). These
constitute a special school that is still little known, with its own
194 M. Bernabei et al. / Journal of Archaeological Science 37 (2010) 192–200

characteristic choice of materials (linings and purflings made of

Table 1
List of analysed stringed instruments from the Collection of the Cherubini Conser-
beech wood, whereby the black part of the purflings was some-
vatory, at the Department of Musical Instruments of the Accademia Gallery, times made from tinted whalebone) and style (very high arching
Florence. that recalls the German school; short and vertically-positioned
Inv. no. Instrument Hypothesized Origin Maker
f-holes that are rather distant one from another). The concentration
date of a representative number of instruments from this school in
1988/035 Cello – Rome David Tecchler a single collection facilitates a series of wide-ranging investiga-
1988/039 Cello – – Anonymous, Tuscan tions, aiming to determine both common and individual charac-
school teristics of these instruments (Rossi Rognoni, 2009). The Collection
bm909 Cello 1968 Rimini Marino Capicchioni
includes, in fact, at least one instrument each of all the principal
1988/044 Double bass 1827 Livorno Giuseppe Bracci
1988/042 Double bass 179? Florence Luigi Piattellini
violin-makers belonging to the golden era of the Tuscan school.
1988/027 Controviolino 1901 Florence Valentino De Zorzi Three violins are attributed to its most prominent representative,
1988/028 Controviolino 1902 Florence Valentino De Zorzi Giovanni Battista Gabbrielli (1716–1771), whose instruments had
1988/029 Controviolino 1904 Florence Valentino De Zorzi already attained international fame in Europe during the life of
1988/030 Controviolino 1904 Florence Valentino De Zorzi
their maker. Of these, only one is signed and labelled (1988/008),
1988/031 Controviolino 1908 Florence Valentino De Zorzi
1988/032 Controviolino 1910 Florence Valentino De Zorzi whilst another, although it does not bear a label, exhibits its mak-
1988/020 Bass-viola 1874 Arezzo Lorenzo Arcangioli er’s brand (1988/009). While, for the third instrument (1988/237),
1988/016 Viola First half Mirecourt Anonymous its dendrochronological analysis can yield useful information
of the 19th regarding its likely attribution.
century
1988/017 Viola – Naples Johannes Gagliano
One viola bears a label of Bartolomeo Bimbi (1988/022), a violin-
1988/018 Viola First half Mirecourt Anonymous maker who worked in Siena and Florence during the second half of
of the 19th the eighteenth century, and whose importance amongst the Flor-
century entine violin-makers is confirmed not only by contemporary critics
1988/021 Viola First half Central Anonymous
of his work, but also by the commission he received from the Court
of the 18th Italy
century of the Grand Duke as controller of all the custom duties for musical
1988/022 Viola 1770 Florence Bartolomeo Bimbi instruments in the region of Tuscany (Rossi Rognoni, 2002).
1988/023 Viola First half Southern Anonymous Two instruments (1988/007 and 1988/026), a violin and a viola,
of the 20th Italy both signed, are attributed to the workshop of Lorenzo and Tommaso
century
1988/024 Viola 1809 Perugia Pietro Pallotta
Carcassi, both of whom had connections with the Lorenese Court,
1988/025 Viola 1915 Florence Serafino Casini where they carried out restoration and maintenance work on almost
1988/026 Viola 1786 Florence Lorenzo and all the stringed instruments of the Collection (Montanari, 1997).
Tommaso Carcassi
2002/312 Viola Early 20th Mirecourt Anonymous
2.2. The video time table
century
2002/313 Viola 1978 Florence Luciano Sderci
2002/314 Viola 1919 Pieve di Gotti, Orsolo The great part of the measurements was carried out by using the
Cento (FE) Video Time Table (VIAS, 2005), an instrument that combines
bm901 Viola 1968 Florence Iginio Sderci a portable measuring device and a digital, high-resolution video
1988/004 Violin – Naples Nicolò, Ferdinando
and Giuseppe
camera (Fig. 2).
Gagliano The device has the following advantages:
1988/005 Violin 1719 Salzburg Joannes Schorn
1988/006 Violin 1722 Salzburg Andreas Ferdinand - The tree rings can be measured on site;
Mayr
- The measurements are not invasive;
1988/007 Violin 1767 Florence Lorenzo and Tommaso
Carcassi - The correctness of the measurements can be checked
1988/008 Violin 1764 Florence Giovanni Battista immediately.
Gabbrielli
1988/009 Violin 1770 – Giovanni Battista The measuring device consists of four fundamental parts:
Gabbrielli
1988/011 Violin – Pesaro Del Coradel
a tripod, the optics, a three-axes movement device and an external
1988/012 Violin 1784 Livorno Antonio Gragnani unit for the movement’s control. The whole system is connected to
1988/013 Violin 1861 Turin Antonio Guadagnini a portable computer that analyses the data. The optics consist of
1988/014 Violin 1886 Florence Giuseppe Scarampella a digital video camera with a focal distance of 20 cm, which avoids
1988/236 Violin – Mittenwald school of Joan
any direct contact with the wooden object (Fig. 2). The three-axes
Carol Kloz
1988/237 Violin – – attr. to Giovanni movement device allows movements with a precision of 1/
Battista Gabbrielli 8000 mm. The tree-ring series obtained can be visualized and
2002/300 Violin – – Tyrolean school processed with the PAST4 software of SCIEM (Scientific Engineering
2002/301 Violin – – German school and Manufacture). The VTT’s video control enables its user to save
2002/302 Violin Early 20th Saxony Anonymous
the most important images, a service that has proved to be very
century.
2002/303 Violin – Germany Anonymous useful in those cases where doubts arose regarding the interpre-
2002/304 Violin 1830 Paris François Breton tation of a sequence.
2002/305 Violin 1920 Milan Leandro Bisiach
2002/306 Violin 1927 Florence Josef Bargelli
2.3. Sampling procedure
2002/307 Violin 1926 Florence Lapo Casini
2002/308 Violin 1978 Florence Lapo Casini
2002/309 Violin 1982 Florence Lapo Casini Basically, four tree-ring series were obtained from the belly of
2002/311 Violin 1977 Florence Luciano Sderci each musical instrument: two each from the bass and treble sides,
bm900 Violin 1967 Bologna Ansaldo Poggi which were repeated at various parts of the belly in order to
maximize the number of growth rings available and, at the same
 M. Bernabei et al. / Journal of Archaeological Science 37 (2010) 192–200
Fig. 2. The dendrochronological measuring device used in the study. In some cases,
dendrochronological analysis was possible without removing the instrument from its case.
time, to avoid errors caused by possible distortions in the veining.
The possibility of immediate comparisons between the dendro-
chronological series measured allowed the repetition of
a measurement whenever anomalies were detected in the
tendency of a ring curve.
The above-described sampling procedure was adapted to the
characteristics of each instrument. In particular, the number of
elements that make up the belly of each instrument was taken into
consideration: one, two or more pieces. Consequently, the number
of measurements was increased or reduced, accordingly.
The tree-ring measurements undertaken by the VTT were
always accompanied by photographs of the wooden surface. These
digital photographs permitted a constant comparison, on the
monitor, between the wooden surface analysed and the dendro-
chronological series recorded. Where the varnish on the instru-
ment belly was not transparent enough for the VTT camera, the
photographic sampling method was used to measure the tree rings.
Tree-species identification was carried out on the belly of the
instruments, in a non-invasive manner, by Marco Fioravanti and
Giovanni Signorini of the University of Florence, Italy.
2.4. Statistical tests
Statistical tests are usually of great help in dendrochronological
dating. However, if used without the necessary discretion, they
have serious deficits. In particular, they may give rise to the
following errors (Sander and Levanic, 1996):
- Wrong dates believed to be correct because they are associated
with high, casual correlation values (type I errors);
195
- Correct dates not accepted because they are associated with
occasional low correlation values (type II errors).
In order to avoid this kind of problem, each tree-ring sequence
was confronted visually and statistically with more than one
reference chronology. The reference chronologies used are those
already published and valid for Alpine and central European
Norway spruce (Topham and McCormick, 1998, 2000; Grissino-
Mayer et al., 2004). The International Tree-Ring Database (http://
hurricane.ncdc.noaa.gov/pls/paleo/fm_createpages.treering) that
lists a large number of high-quality and well-dated chronologies
from all over the world, was a great help. Dates were considered
reliable only after they had been confirmed by more than one
reference chronology.
The statistical cross-dating tests used in this study are:
- t-values: adapted to time-series by Baillie and Pilcher (1973).
- Gleichläufigkeit (Glk): a measure of the year-to-year agreement
of the ring-growth tendencies of two chronologies, expressed
as a percentage of cases of agreement from one year to the next
(Kaennel and Schweingruber, 1995).
- Statistical significance of Glk: can be at 95.0%, 99.0% or at 99.9%
and has been indicated here as *, ** and ***, respectively.
Sequences that cross-matched with t-values >4, and with cor-
responding high values of Gleichläufigkeit and high statistical
significance values in more than one reference chronology, were
considered to be statistically reliable.
CATRAS and PAST4 computer programmes were used for visu-
alizing the series and for carrying out statistical synchronization
tests. Considering the variability of t-values in relation to the
software employed (Sander and Levanic, 1996), and in order to keep
the following tables uniform, it was decided to adopt the values
calculated by the PAST4 programme in this study.
3. Results
A total of forty-nine instruments was dendrochronologically
analysed, 192 chronologies were built and more than ten thousand
year rings were measured. Where more than one measurement had
been obtained from the same instrument belly, a comparison of the
sequences always showed high correlation values (mean t-val-
ue > 11). For this reason, a representative mean chronology was
built for each instrument whose belly consisted of more than one
piece (Table 2). In the case of four instruments, no tree-ring
measurement was possible, whereas in five cases only one belly
piece was measurable, although the instrument’s belly contained
more than one piece. Here, the rings were not clearly visible
because the varnish was not transparent enough, in addition to
being obscured by the patina of many years. The photographic
sampling method was, therefore, used for three instruments (1988/
023, 1988/026 and 2002/305), which, however, did not yield
correlation values high enough for dating.
Thirty-seven instruments (75%) were successfully dated (Fig. 3).
The remaining eight (twelve minus the four unmeasured ones)
were considered undatable because of an insufficient number of
identifiable rings (<50), or because the statistical correlation values
were not high enough (t-values < 4). Generally, however, the
correlation values of cross-dating individual instrument curves
against the reference chronologies were high (mean t-value: 6.08
and Glk 69.27, with a statistical significance of 95% or higher; in
twenty-eight cases as high as 99.9%). There were particularly high
correlations with the Alpine chronologies of Siebenlist-Kerner
(1984) and Hüsken and Schirmer (1993).
196 M. Bernabei et al. / Journal of Archaeological Science 37 (2010) 192–200

Table 2
The belly: number of tree rings in each instrument’s series, number of measurements for each belly piece, arrangement of the pieces (B being the bass side, T the treble side and
Dt is the difference between each instrument’s label date and its terminus post quem date).
Inv. no. Instrument Tree rings Measurements/pieces Arrangement Hypothesized Dendrochronol. Dt
of pieces date date
1988/035 Cello 108 2/2 /) – 1699 –
1988/039 Cello 109 2/2 /) – 1812 –
bm909 Cello 115 2/2 /) 1968 1952 16
1988/042 Double bass 130 2/2 /) 179? 1781 14 (?)
1988/044 Double bass 209 1/8 – 1827 – –
1988/027 Controviolino 62 2/4 – /) – 1901 1780 121
1988/028 Controviolino 55 1/2 /? 1902 1814 88
1988/029 Controviolino 87 2/2 )/ 1904 1793 111
1988/030 Controviolino 89 2/4 – /) – 1904 1791 113
1988/031 Controviolino 79 2/2 /) 1908 1756 152
1988/032 Controviolino 99 2/2 /) 1910 1726 184
1988/020 Bass-viola 114 2/2 /) 1874 1847 27
1988/016 Viola 139 2/2 /) First half of the 19th century 1810 –
1988/017 Viola 107 2/2 /) – 1765 –
1988/018 Viola 169 1/1 B/T First half of the 19th century 1816 –
1988/021 Viola 95 2/2 )) First half of the 18th century 1671 >29
1988/022 Viola 99 2/2 /) 1770 1761 9
1988/023 Viola – 0/2 /) First half of the 20th century – –
1988/024 Viola 95 2/2 /) 1809 1796 13
1988/025 Viola 59 1/2 )) 1915 – –
1988/026 Viola – 0/? ? 1786 – –
2002/312 Viola 91 2/2 /) Early 20th century 1880 >20
2002/313 Viola 25 1/5 – 1978 – –
2002/314 Viola 80 2/2 /) 1919 1904 15
bm901 Viola 88 2/2 /) 1968 1953 15
1988/004 Violin 53 2/2 /) – 1712 –
1988/005 Violin 72 2/2 /) 1719 – –
1988/006 Violin 52 2/2 /) 1722 – –
1988/007 Violin 90 2/2 /) 1767 1743 24
1988/008 Violin 110 2/2 /) 1764 1754 10
1988/009 Violin 58 2/2 /) 1770 1768 2
1988/011 Violin 202 1/1 B/T – 1658 –
1988/012 Violin 94 2/2 /) 1784 – –
1988/013 Violin 148 1/1 B/T 1861 1803 58
1988/014 Violin 80 2/2 /) 1886 1859 27
1988/236 Violin 98 2/2 /) – 1729 –
1988/237 Violin 92 2/2 /) – 1749 –
2002/300 Violin 116 1/1 B/T – 1749 –
2002/301 Violin 127 2/2 /) – 1746 –
2002/302 Violin 136 2/2 /) Early 20th century – –
2002/303 Violin 64 2/2 /) – 1771 –
2002/304 Violin 303 1/1 B/T 1830 1698 132
2002/305 Violin – 0/? ? 1920 – –
2002/306 Violin 70 2/2 /) 1927 – –
2002/307 Violin 82 2/2 /) 1926 – –
2002/308 Violin 72 2/2 /) 1978 1906 72
2002/309 Violin – 0/2 /) 1982 – –
2002/311 Violin 91 2/2 /) 1977 1944 33
Bm900 Violin 80 2/2 /) 1967 1940 27

For each dated sample, a value Dt was calculated, which is
defined as the difference between each instrument’s label date and
its terminus post quem date (Tables 2, 3).
The mean chronology built from the tree-ring series of the
individual instruments, which was called Accademia Master
Chronology (AMC01), consists of 558 year rings (Fig. 3) dating from
1396 to 1953, and it correlates very well with other Alpine Norway
spruce reference chronologies. Furthermore, there is very good
cross-matching with central European spruce reference chronolo-
gies (high t-values, usually above 5), but also with master chro-
nologies from different species, such as silver fir (Abies alba Mill.)
and larch (Larix decidua Mill.) (Table 4). Strangely enough, the
sequences of some instruments that correlate with the spruce
master chronologies, cross-match better with the reference chro-
nologies of other species (Table 5).
Some instrument tree-ring series cross-match very well
amongst themselves. In particular, the wood used by Gabbrielli,
Bimbi, Carcassi and Guadagnini shows the highest correlation
values, with mean t-values of 8.25.
Finally, for each instrument category, the mean tree-ring width
and standard deviation were calculated (Table 7). Although some
categories are represented by only a few instruments, which is the
case for bass-viola and double bass, others contain enough
instruments to render the data significant.
4. Discussion
In the light of the results obtained (Table 2), the greater part of
instrument bellies were built following the usual method,
mentioned in the Introduction, which foresees the splitting in half
of a single, quarter-sawn, wedge and joining the two halves where
the youngest growth rings converge (arrangement /)).
There are, however, also instruments (violas 1988/025, 1988/
021), in which the arrangement of the belly pieces, although they
 M. Bernabei et al. / Journal of Archaeological Science 37 (2010) 192–200 197

Fig. 3. Dendrochronological dates of the stringed instruments from the Cherubini Collection.

are from the same wedge, is such that the rings grow in the same construction of 22 instrument bellies came from the same tree (cfr.
direction (arrangement ))); or instrument bellies (controviolino Topham and McCormick, 2000).
1988/029) where the direction of growth is oriented towards the Sometimes, the dendrochronological series obtained from two
outer edge (arrangement )/). Other exception are instruments pieces of the same instrument belly are slightly offset in their
whose belly is made from a single piece, in which the growth respective dating; the ‘‘missing’’ rings probably correspond to the
direction always points from the bass to the treble side, as well as piece of wood used to make the bass-bar: a small fillet that was
those made from more than two pieces: a double bass (1988/044) carefully shaped to adhere perfectly to the internal curvature of the
and two controviolini (1988/027, 1988/030). The latter instruments case, and which assists in the equal distribution of pressure and
sometimes contain very small additional pieces of wood, at times vibrations throughout the body.
placed towards the outside near the purfling, whose purpose it is to 75% of the instruments were successfully dated, thereby deter-
give the instrument the required shape and dimension. These small mining the terminus post quem date of manufacture. Compared
additional parts were not analysed dendrochronologically because with similar research work of the past (Klein et al., 1986; Topham
they contain an insufficient number of rings. Finally, there is one and McCormick, 1998, 2000), our percentage of success is some-
viola (2002/313) whose belly consists of five pieces, all with an what higher. The reason for this probably lies in the homogeneity of
insufficient number of rings (between sixteen and twenty-five), the analysed wood, which mainly derives from central and
whose separate elements during a visual examination showed northern Italy, thereby giving rise to high statistical values of
great similarities of ring pattern. synchronization between individual instrument ring-curves and
In all thirty-six instruments whose belly consists of more than the master chronologies. In this context, the works of Gabbrielli,
one piece, the tree-ring sequences of the various pieces showed Bimbi, Carcassi and Guadagnini must be mentioned, which are all
a high degree of cross-matching, with a mean t-value of 11.75
(varying between 4.02 and 28.20). In twenty-two of them, the t- Table 4
value was >10. Hillam and Groves (1996) used this threshold to Cross-matching the Accademia Master Chronology against some reference chro-
indicate that oak samples with a larger t-value derive from the nologies valid for the study area (Data downloaded from the International Tree-Ring
Data-Bank, http://hurricane.ncdc.noaa.gov/pls/paleo/fm_createpages.treering).
same tree trunk. Although conifers usually give higher t-values, it
can be assumed that this threshold also applies to Norway spruce. Chronology author Species Site name t-value Glk
Consequently, we can say that the wood employed in the Siebenlist-Kerner Picea abies Karst. Ötztal 14.40 70.90***
Hüsken and Schirmer Larix decidua Mill. Fodara Vedla 9.36 63.50***
Schweingruber Picea abies Karst. Obersaxen 9.26 67.50***
Table 3 Hüsken and Schirmer Picea abies Karst. Fodara Vedla 8.16 64.00***
The mean difference between label date and terminus post quem date (Dt) for Siebenlist-Kerner Larix decidua Mill. Ötztal 7.07 61.80***
instruments produced in the course of three centuries. Schweingruber Picea abies Karst. Cortina 6.98 64.80***
Bigler Picea abies Karst. Davos 6.48 65.00***
Century Mean Dt (years) Min. value (years) Max. value (years) Becker Picea abies Karst. Bayerischer Wald 6.07 63.60***
18th 11.3 2 24 Siebenlist-Kerner Pinus cembra L. Ötztal 5.63 57.50***
19th 51.4 13 132 Hüsken and Schirmer Pinus cembra L. Fodara Vedla 5.17 58.00***
20th 73.9 14 184 Becker Abies alba Mill. Bayerischer Wald 4.98 57.00**
198 M. Bernabei et al. / Journal of Archaeological Science 37 (2010) 192–200

Table 5 Although a proper analysis of individual instruments, with

Cross-matching the De Zorzi instruments’ master chronology against some silver fir a view to the provenance of their wood, has not been carried out,
and spruce reference chronologies.
some details were obtained concerning the origin of the wood.
Chronology author Species Site t-value Glk
Becker Giertz-Siebenlist Abies alba Mill. Tyrol 10.00 71.40*** - Generally, high statistical correlations between the AMC01
Becker Pratomagno, 8.30 64.20*** (Table 4) and the Siebenlist-Kerner (1984) chronology in
Schweingruber Picea abies Karst. Cortina 4.80 58.90**
particular, indicate that a large part of the timber originates
Becker Bayerischer Wald 4.71 60.80***
from the Eastern Alps. Apart from the statistical evidence, this
is confirmed by the opinion of various authors (Henley, 1973;
Harvey, 1995; Corona, 1998), who state that that particular
very similar from a dendrochronological point of view, indicating
region traditionally was one of the most important timber
the same provenance of the wood used for their manufacture. In
supply centres in Europe. Furthermore, the use of Norway
particular, the wood used to make an instrument of uncertain
spruce of Alpine provenance is documented amongst violin-
attribution, inv. no. 1988/237, bears a strong affinity to the one
makers of Nuremberg in Bavaria, another centre that is well-
signed by Giovanni Battista Gabbrielli. This, together with similar-
known for the production of resonance spruce wood (Klein
ities of style, removes all doubts regarding the making of this
et al., 1986), and even amongst British violin-makers (Topham
instrument by this famous violin-maker.
and McCormick, 1998). Hence, when the timber is analysed
Finally, a t-value of 16.40 between the controviolini 1988/029
from a geographical point of view, an Alpine provenance for the
and 1988/030, both attributed to the violin-maker Valentino De
instruments of the Cherubini Conservatory would appear to be
Zorzi, as well as the significant visual comparison (Fig. 4), demon-
most likely. An even more precise geographical location has
strate the use of wood deriving from the same tree for the manu-
been proposed by Corona (1981) who, on the basis of
facture of two different instruments.
dendrochronological considerations and evaluations, demon-
Apart from being entirely consistent with the attributed date of
strated that the wood of the Bimbi viola, which is also subject
manufacture, a comparison between the dendrochronological
of this study, originates from the Val di Fiemme in the Trentino
dates and the label dates on the instruments has shed some light
region, Italy. And Bimbi was one of those violin-makers who
on the working methods of these violin-makers, of whom very
used wood with very similar ring patterns.
little is known besides their surviving instruments. In fact, the
- Some instruments have high correlation values also with other,
interval between the youngest ring and the date of manufacture,
more central European, reference chronologies, for example
called Dt (Table 2), depends on the amount of wood removed
from Germany (Falkenstein, Bavaria) or Switzerland (Ober-
during the construction of the instrument, but it also includes the
saxen). This is true particularly for more recently built instru-
period of wood seasoning, and it gives an indication of how much
ments, from the nineteenth and twentieth centuries. The
time had passed after felling the tree before the wood was used.
instruments with inventory numbers bm900, 2002/311, bm909
Furthermore, all modern violin-makers agree that, traditionally,
and bm901, for example, have higher t-values against the
only a few rings near the bark were removed during the manu-
Bavarian chronology than against the Siebenlist-Kerner master
facture of an instrument and that, if a piece of timber was larger
chronology (mean t-value >5 and >3, respectively). In any case,
than necessary, the inner (older) part was removed, not the outer
the values of these statistical synchronization tests are not high
one (Topham, 2003).
enough to securely attribute the provenance of these instru-
Hence, the value Dt can provide an indirect estimate of the
ments that, however, reach t-values that vary between 7.37 and
seasoning time before manufacture, which is an important tech-
11.90 against the AMC01 master chronology.
nical aspect for violin-makers.
- By contrast, one aspect that is very difficult to understand
Similarly to earlier findings from a study of seventy-two instru-
concerns the correlations between some individual instrument
ments attributed to Antonio Stradivari (Topham, 2003), in our study
tree-ring series and silver fir master chronologies. These simi-
the value Dt varies between a minimum of two and a maximum of
larities have already been observed in the past (Klein et al.,
twenty-four rings for instruments built during the eighteenth
1986). In our case, for example, the mean chronology of De
century (Table 3). For instruments made around the mid-nineteenth
Zorzi’s controviolini (Table 5) corresponds better with silver fir
century, instead, there is a wider range of Dt, here between
than with Norway spruce reference chronologies. This is all the
a minimum of thirteen and a maximum of 132 years. This tendency
more surprising if one considers that the two silver fir chro-
continues during the twentieth century, when the use of old wood,
nologies are very distant one from another, deriving from the
obtained from artefacts, became more frequent, as in the case of six
Bavarian Alps (Becker and Giertz-Siebenlist, 1970) and Tuscany
controviolini of the Collection (with a minimum of 88 to a maximum
(Becker, Italy 022, data from http://hurricane.ncdc.noaa.gov/pls/
of 184 years of difference between the dendrochronological date
paleo/ftpsearch.treering), respectively. One explanation for this
and the date of manufacture) and of a violin by Lapo Casini (72 years
very strange behaviour regards the altitude the wood of some
of difference).

Fig. 4. Two tree-ring curves from controviolini 1988/029 and 1988/030, made by Valentino De Zorzi, stacked one above the other for visual comparison.
 M. Bernabei et al. / Journal of Archaeological Science 37 (2010) 192–200
instruments came from (Wilson and Hopfmueller, 2001). A few
authors (Klein et al., 1986; Grissino-Mayer et al., 2004) have
observed that the timber used in the manufacture of musical
instruments either came from high-altitude locations or from
low altitudes. In our case, efforts of cross-matching with various
low-altitude spruce chronologies, from Bavaria, Switzerland,
France and Italy, have not yielded significant results that could
have answered the question. A comparison has also been made
with the tree-ring series of Stradivari’s famous Messiah violin,
whose wood allegedly came from the low-lying area of the pre-
Alps above Cremona (Grissino-Mayer et al., 2004), but there was
no correlation (t-value ¼ 0.26). Consequently, at the moment
there is no solution for the problem, and further investigations
are needed. Fortunately, instrument sequences showing high
degrees of cross-matching against silver fir chronologies also
correlate, albeit to a lesser degree, with Norway spruce, which
has made it possible to date them.
AMC01 has important potential as a dating tool. This is shown by
comparing this master chronology with the individual tree-ring
series it consists of (each time removing from the master the
individual curve that is to be compared), and also by comparing it
with the sequences of two important instruments from the Royal
Academy of Music in London: the violas Archinto and Kux-Cas-
telbarco (Grissino-Mayer et al., 2004; Table 6), which have shown
high degrees of cross-matching with AMC01. The reason probably
lies in the characteristics of the wood employed in the manufacture
of the instrument bellies. In fact, as already mentioned above,
spruce wood that is to be used for making instruments must be free
from defects, including reaction wood, resin pockets, knots and so
on; it must be without growth anomalies and should have a regular
ring pattern with reduced tree-ring widths (Burckle and Grissino-
Mayer, 2003; Bucur, 2006). Apart from this, the wood must also
have the right technological properties, such as a low ratio between
wood density and modulus of elasticity – characteristics that are
fulfilled by only very few trees whose wood is, therefore, suitable
for making musical instruments (Ono and Norimoto, 1983; Bucur,
2006; Buksnowitz et al., 2007).
In the greater part of the instruments analysed, such charac-
teristics have been associated with tree growth at high altitudes,
which is confirmed by the high correlation values of their curves
against Alpine chronologies.
All this is reflected in a general conformity of growth increment,
which makes it easy to come across extraordinarily high correlation
values for individual instrument curves, when compared with the
cross-matches usually found in dendrochronological dating. The
AMC01 reflects the growth increments of thirty-seven tree-ring
series with the above-mentioned characteristics; consequently, it
contains a very homogeneous signal, which is useful in compari-
sons with wood of similar qualities (Table 6).
Finally, the relationship between tree-ring widths and category
of musical instruments should be mentioned.
In the past, the wood employed in the manufacture of musical
instruments was chosen on the basis of visual characteristics and by
‘‘feeling’’ it; for many craftsmen this is still the case today. One of
Table 6
Cross-correlation of the viole Archinto and Kux-Castelbarco (Grissino-Mayer et al.,
2004) with the Siebenlist-Kerner and the AMC01 chronologies (Data from the
International Tree-Ring Data-Bank, http://hurricane.ncdc.noaa.gov/pls/paleo/fm_
createpages.treering).
Instrument Siebenlist-Kerner AMC01
t-value Glk t-value Glk
Viola Archinto 2.38 52.50 5.31 64.20***
Viola Castelbarco 4.83 65.20*** 6.32 64.80***
199
Table 7
The mean ring widths of different categories of instruments (The data for double
bass and bass-viola are not representative as only one instrument each was
analysed.).
Instrument category No. of Mean ring width SD
instruments (mm/100) (mm/100)
Violin 23 123.43 35.84
Viola 14 135.29 34.88
Bass-viola 1 116.18 53.93
Controviolino 6 170.36 33.88
Cello 3 191.53 50.26
Double bass 1 215.18 83.27
the most important characteristics of the wood is veining, which
refers to the width and regularity of rings.
Table 7 shows the mean tree-ring width and standard deviation
for each instrument category. Although it is known that the ring
widths on the belly of an instrument affect the instrument’s
acoustic properties (Blossfeld et al., 1962; Buksnowitz, 2006;
Bucur, 2006), Beuting and Klein (in Buksnowitz, 2006), who ana-
lysed 594 instruments, were unable to find significant variations
between the relative growth increments of various instrument
categories. By contrast, in the Cherubini Collection, the average
tree-rings width of violins are smaller than those of other stringed
instruments; in fact, they increase in proportion to instrument size
and are widest in cello and double bass, whereby large rings
correspond to bass tones. This relationship has been confirmed by
other research work (Blossfeld et al., 1962; Feuerstein, 1935; Zim-
mermann, 1996) and it would seem to follow a theoretic principle
whereby, in instruments of the same size, wood with narrow rings
vibrates for a longer time and in higher tones (Ille, 1976), whilst
wood with wide rings emphasizes deeper sounds (Blossfeld et al.,
1962; Bariska, 1996).
Standard deviation indicates the spread of data around the
mean, and it may be considered an index of the regularity of growth
rings. Although it is invariably influenced by the number of
samples, the standard deviation for each instrument category is
rather homogeneous, at least as far as small instruments are con-
cerned. In general, it has been observed (Zieger, 1960) that wood
with sudden variations in the ring pattern is not suitable for
building musical instruments because this wood has less homo-
geneous acoustic properties. A higher value for the standard devi-
ation of instruments such as cello and double bass could simply be
due to the greater quantity of timber required for their
manufacture.
In summary, analysing the data in Table 7, we can see that Italian
and, in particular, Tuscan violin-makers carefully chose their timber
with regard to its veining, which has favoured narrow rings in small
instruments with higher tones and has prevented great variations
in the regularity of growth increments.
5. Conclusions
This study is part of a wider range of research work aiming to
improve the possibilities of comparing and the general knowledge
of stringed musical instruments from the Cherubini Conservatory.
The dendrochronological analysis has supplied useful information
regarding their date of manufacture and attribution.
Right from the start, the dendrochronological investigations
have been based on a continual exchange of data and information
between research workers of different disciplines. The raw data of
the dendrochronological series were shared equally, and the elab-
oration of the results was described and discussed, thereby deter-
mining a reliable terminus post quem date for each instrument.
200 M. Bernabei et al. / Journal of Archaeological Science 37 (2010) 192–200
The analyses have shown that, with a few important exceptions,
the bellies of most instruments were constructed following tradi-
tional techniques, and that a great part of the Norway spruce timber
used derives from the Eastern Alps, although there are some
affinities with more continental reference chronologies. A mean
chronology, called Accademia Master Chronology (AMC01), was
successfully built up and correlates very well with Alpine and
central European reference chronologies. AMC01 also has impor-
tant potential as a dating tool for musical instruments. Furthermore,
the wood of the instruments analysed indicates a relationship
between tree-ring width and acoustic properties of the various
categories of instruments.
This work demonstrates that our understanding of the musical
instruments production benefits from the intensive collaboration
between researchers from different disciplines. The interaction
between musical instruments experts and dendrochronologists
proved extremely fruitful and resulted in a more accurate inter-
pretation of the acquired data from both disciplines.
Acknowledgements
The authors thank Christa E. Backmeroff for her helpful
comments and assistance during the manuscript preparation. We
want also thank Claudio Pollini for his suggestions and practical
contribution.
References
Baillie, M.G.L., 1982. Tree-Ring Dating and Archaeology. The University of Chicago
Press Ltd, London.
Baillie, M.G.L., Pilcher, J.R., 1973. A simple cross-dating program for tree-ring
research. Tree-Ring Bulletin 33, 7–14.
Bariska, M., 1996. Zur Geschichte der Holzverwendung beim Musikinstru-
mentenbau. Schweizerische Zeitschrift für Forstwesen 147 (9), 683–693.
Becker, B., Giertz-Siebenlist, V., 1970. Eine über 1100jaehrige mitteleuropäische
Tannenchronologie. Flora 159, 310–346.
Bernabei, M., Quarta, G., Calcagnile, L., Macchioni, N., 2007. Dating and technolog-
ical features of wooden panel painting attributed to Cesare da Sesto. Journal of
Cultural Heritage 8, 202–208.
Blossfeld, O., Haasemann, W., Haller, K., 1962. Klangholz und Klangholzsortierung.
Deutsche Sozialistische Forstwirtschaft 12, 140–145.
Bucur, V., 2006. Acoustics of Wood. Springer Series in Wood Sciences, second ed.
Springer-Verlag, Berlin, ISBN 3540261230.
Buksnowitz, C., 2006. Resonance wood of Picea abies. Doctoral thesis, Institute of
Wood Science and Technology, BOKU, Vienna.
Buksnowitz, C., Teischinger, A., Müller, U., Pahler, A., Evans, R., 2007. Resonance wood
(Picea abies (L.) Karst.) – evaluation and prediction of violin makers’ quality-
grading. Journal of the Acoustical Society of America 121 (4), 2384–2395.
Burckle, L., Grissino-Mayer, H.D., 2003. Stradivari, violins, tree rings, and the
maunder minimum: a hypothesis. Dendrochronologia 21 (1), 41–45.
Corona, E., 1980. Ricerche dendrocronologiche su due violini del XVIII secolo. L’Italia
Forestale e Montana 35, 112–115.
Corona, E., 1981. La viola Bimbi ha ‘‘ascendenze’’ trentine. Natura Alpina 32, 27–29.
Corona, E., 1998. Caratterizzazione dendrocronologica degli strumenti liutari. Legno
Cellulosa Carta 1, 16–20.
Falletti, F., Meucci, R., Rossi Rognoni, G., 2001. La musica e i suoi strumenti. Giunti,
Firenze.
Feuerstein, A., 1935. Das Klangholz. Forstwissenschaftliches Centralblatt 57 (19),
617–624.
Fritts, H.C., 1976. Tree Rings and Climate. Academic Press, London.
Grissino-Mayer, H.D., Deweese, G.G., 2005. Tree-ring dating of the Karr-Kousse-
vitzky double bass. A case of study in dendromusicology. Tree-Ring Research 61
(2), 77–86.
Grissino-Mayer, H.D., Sheppard, P.R., Cleaveland, M.K., 2004. A dendroarchaeo-
logical re-examination of the ‘‘Messiah’’ violin and other instruments attributed
to Antonio Stradivari. Journal of Archaeological Science 31, 167–174.
Harvey, B.W., 1995. Violin Fraud: Deception, Forgery, Theft and the Law. Clarendon
Press, Oxford.
Henley, W., 1973. Universal Dictionary of Violin and Bowmakers. Amati Publishing
Company, Brighton.
Hillam, J., Groves, C., 1996. Tree ring research at Windsor Castle: aims and initial
results. In: Dean, G.F., Meko, D.M., Swetman, T.W. (Eds.), Tree Rings, Environ-
ment and Humanity. Radiocarbon. University of Arizona, Tucson, p. 515.
Hüsken, W., Schirmer, W., 1993. Drei Jahrringchronologien aus den Pragser Dolo-
miten/Südtirol. Dendrochronologia. 11, 123–137. Data received from World Data
Center for Paleoclimatology, Boulder, Colorado, USA.
Ille, R., 1976. Eigenschaften und Verarbeitung von Fichtenresonanzholz für Meis-
tergeigen (II). Holztechnologie 17, 32–35.
Kaennel, M., Schweingruber, F.H., 1995. Multilingual Glossary of Dendrochronology.
WSL/FNP, Birmensdorf. Haupt Pub., Berne, Stuttgart, Vienna, ISBN 3-258-
05259-X, pp. 130–131.
Klein, P., 1985. Dendrochronologische Untersuchungen an Gemäldetafeln und
Musikinstrumenten. Dendrochronologia 3, 25–44.
Klein, P., Mehringer, H., Bauch, J., 1984. Tree-ring chronology of spruce wood and its
application in the dating of stringed instruments. In: Proceedings, ICOM,
Committee for Conservation, Seventh Triennial Meeting, September 1984,
Copenhagen, pp. 69–72.
Klein, P., Mehringer, H., Bauch, J., 1986. Dendrochronological and wood biological
investigations on string instruments. Holzforschung 40, 197–203.
Lottermoser, W., Meyer, J., 1958. Über die Möglichkeit einer Dendrochronologie von
altitalianischen Geigen. Instrumentenbauzeitschrift 12, 295–296.
Montanari, G., 1997. Conservazione e restauro degli strumenti ad arco alla corte di
Firenze in epoca lorenese (1737–1770). In: Meucci, R. (Ed.), Liuteria, Musica e
Cultura, pp. 3–19. Lim, Lucca.
Munro, M.A.R., 1984. An improved algorithm for cross-dating tree ring series. Tree-
Ring Bulletin 44, 17–27.
Ono, T., Norimoto, M., 1983. Study on Young’s modulus and internal friction of wood
in relation to the evaluation of wood for musical instruments. Japanese Journal
of Applied Physics 22, 611–614.
Rossi Rognoni, G., 2002. Le botteghe fiorentine di strumenti musicali. In: Spinelli, R.
(Ed.), Arti Fiorentine: la grande storia dell’artigianato. Il Seicento e il Settecento,
vol. V. Giunti, Firenze, pp. 133–149.
Rossi Rognoni, G., 2004. Alcune fonti sulla produzione liutaria in Toscana nel XIX
secolo. In: Liuteria in Toscana: i liutai del Novecento, Cremona, Cremonabooks,
pp. 33–45.
Rossi Rognoni, G., 2009. Gli strumenti ad arco: Collezione del Conservatorio di
Musica di Firenze. Sillabe, Livorno.
Sander, C., Levanic, T., 1996. Comparison of t-values calculated in different
dendrochronological programmes. Dendrochronologia 14, 269–272.
Siebenlist-Kerner, V., 1984. Der Aufbau von Jahrringchronologien für Zirbelkiefer,
Lärche, und Fichte eines alpinen Hochgebirgsstandortes. Dendrochronologia 2,
9–29.
Topham, J., 2003. A dendrochronological study of violins made by Antonio Stradi-
vari. Journal of the American Musical Instrument Society 29, 72–96.
Topham, J., McCormick, D., 1998. A Dendrochronological investigation of British
stringed instruments of the violin family. Journal of Archaeological Science 25,
1149–1157.
Topham, J., McCormick, D., 2000. A dendrochronological investigation of stringed
instruments of the cremonese school (1666–1757) including ‘‘The Messiah’’
violin attributed to Antonio Stradivari. Journal of Archaeological Science 27,
183–192.
UNI Standard 11141, 2004. Wood dendrochronological dating guidelines. Ente
Nazionale Italiano di Unificazione, Milano.
V.I.A.S., Vienna Institute of Archaeological Science, 2005. Video Time Table. Instal-
lation and instruction manual. Rev. 2.1, Vienna.
Wegst, U., 2006. Wood for sound. American Journal of Botany 93, 1439–1448.
Wilson, R.J.S., Hopfmueller, M., 2001. Dendrochronological investigations of Norway
spruce along an elevational transect in the Bavarian Forest, Germany. Den-
drochronologia 19, 67–79.
Zieger, E., 1960. Untersuchungen über äußere Merkmale, Holzeigenschaften und
forstgeographische Vorkommen der Resonanzqualitäten bei Fichte und einigen
anderen Holzarten. Mitteilungen aus der Staatsforstverwaltung Bayerns 31,
285–298.
Zimmermann, U., 1996. Anforderungen an das Klangholz. Schweizerische Zeitschrift
für Forstwesen 147 (9), 695–702.