SECRETS OF THE ‘MESSIAH’

In 2016 the ‘Messiah’ Stradivari was the subject of an extensive CT scanning project. Francesco Piasentini and Gregg Alf examine the resulting data, discovering repair work in the neck, and attempt to determine how it had originally been set

The date was Monday 5 September 2016 and Francesco Piasentini, a violin maker with a PhD in metallurgical engineering, was waiting at TEC Eurolab, a leading company in non-destructive testing near Modena, for the return of the ‘Messiah’. At 11am Gregg Alf, a fellow violin maker and expert from Cremona’s Museo del Violino (MdV), would be delivering one of the rarest violins on earth. Accompanied by Fausto Cacciatori and Colin Harrison, curators from the MdV and Oxford’s Ashmolean Museum respectively, Antonio Stradivari’s iconic 1716 violin was returning to the city of its birth in celebration of its 300th anniversary.

With assistance from the Ashmolean, Cacciatori and Alf, aided by a team of experts from the MdV’s Arvedi Laboratory of Non-Invasive Diagnostics, had devised a programme of scientific inquiry that would respond, point by point, to questions left unanswered by experts before them. On the day of the violin’s arrival in Italy, its first stop would be a few hours at TEC Eurolab for a series of industrial computed tomography (CT) scans.

X-rays of the past provided snapshots of physical conditions invisible to the naked eye. But CT scanning, made by merging thousands of X-ray-like images into a three-dimensional volume viewed with special software, allows an unprecedented view of conditions below the surface. Directed by Alf’s keen awareness of the issues and guided by Piasentini, who as a CT consultant has developed several applications for the study of bowed stringed instruments (with the support of TEC Eurolab’s team), today’s scanning would hopefully provide useful information for a worldwide community of violin makers.

As the first radiographic images of the ‘Messiah’ began appearing on the monitor (figure 1), it was clear that the instrument presented unique and unexpected features. Of particular interest were remnants of old iron nails still present in the neck heel, and a single metal screw that appeared to be the only attachment now securing the violin’s neck to its non-mortised body.

Gregg Alf preparing the ‘Messiah’ for scanning at TEC Eurolab

PHOTOS ROBERTO DOMENICHINI @ MUSEO DEL VIOLINO. CT SCANS FRANCESCO PIASENTINI @ TEC EUROLAB

FIGURE 1 The first images of the ‘Messiah’ under the X-ray. A couple of nail fragments are still present in the neck heel. A single screw fastens the neck to the body.

A first scan of the complete instrument was performed at 110 microns, and was immediately reconstructed using NSI efX-CT software. The C-bout region was then re-scanned at a higher resolution (down to 73 microns), nearly doubling the resolution in the soundpost area, which would prove to be crucial in retracing the history of the instrument.

The presence of metals in the neck/body joint region required a third scan. Metals are a primary source of artefacts when scanning violins. Their presence requires ‘harder’ X-rays, obtained by increasing the imaging energy and filtering out the lower ‘soft’ part of their spectrum. A third scan was therefore set up with higher voltage and resolution, and was limited to the neck/body joint region.

In the end we made three scans of the violin, and a fourth for its fittings. In this article we will focus only on the complete violin and the neck/body scans. These scans were aligned and merged to present the best information from both. Piasentini has been using industrial CT for violins since 2014 and, like Alf, has studied many examples of fine Italian makers. Because of its pristine condition, the ‘Messiah’ remains one of the most interesting and informative instruments of all. As the first CT images of the ‘Messiah’ became visible, a question occurred: without touching the violin and working only on its CT datasets, would it be possible to re-set the neck,virtually, to its original position? A realignment of the original nail holes with the nails that caused them 300 years ago would allow us to confirm key aspects of Stradivari’s original Baroque set-up. Piasentini performed a first ‘virtual neck reset’ in 2018, showing promising results. Over the following months he worked with Alf and Cacciatori to clarify and refine the conclusions presented in this article.

Francesco Piasentini at the workstation with a preview of the results of the first scan

A CT ‘volume’ is basically a stack of digital images called CT ‘slices’. Most violin makers are familiar with CT slices of instrument bodies and scrolls from examples of high-resolution industrial CT scanning that have appeared in past editions of The Strad. The simplest way to explore a CT volume is to scroll through its slices. Scrolling through the top-block slices of the ‘Messiah’ was like uncovering the various strata at an archaeological dig. As with archaeology, by interpreting what we see, we can reconstruct a plausible history. Starting from the upper ribs and moving in towards the centre of the instrument, we travelled through the top-block of the ‘Messiah’. It is made from two separate sections of willow: an original section glued to the ribs and a second section glued to the first section during a subsequent restoration.

In the original section (figure 2), seven holes are present. We presume this to be Stradivari’s original top-block because the small alignment pin between the top-block and back plate is still intact and visible in our scan. Four of the holes comprise the characteristic ‘diamond’ pattern we find in many instruments that still retain their original top-block, such as the 1690 ‘Tuscan-Medici’, the 1693 ‘Harrison’ and the 1713 ‘Huberman’. The other three are arranged in a pattern similar to what we see in a surviving original Stradivari neck now archived at the MdV (MS128).

After a thin layer of glue, we come to a second section of the top-block shown in figure 3. It is clearly from a later addition and made from a different piece of wood. In this section only the three nail holes are present. These holes were closed with wooden nails, but are clearly visible inside the instrument using a dental mirror, as is a shiny metal screw.

The three holes can therefore be described as a second nailing, made after the original ‘four nails’ neck setting had been detached. The absence in this section of any trace of the four holes in a diamond pattern reinforces this assumption.

Judging from the top-block, we may conclude that the ‘Messiah’ neck was nailed at least two times. The first was with four nails, arranged in the diamond pattern. The second had three nails. Additional proof comes from the analysis of the neck heel. When we look at the original part of the neck, we still find traces of the four-nail and three-nail patterns seen in the top-block, but with one important distinction. Nail fragments are still present in the neck heel, but only to those holes belonging to the original, diamond-shaped, four-nail pattern. No metal fragments from the three-nail pattern are present. Instead we see only their holes.

THE ‘MESSIAH’ NECK WAS NAILED AT LEAST TWO TIMES, THE FIRST WITH FOUR NAILS, THE SECOND WITH THREE

FIGURE 2 A CT slice of the original part of the top-block region, showing the nail holes and the central screw (note the small wooden top and back alignment pins, still intact). Red arrows point to the original four-nail pattern. Green arrows point to the three-nail pattern.

FIGURE 3 CT slice of the replaced part of the upper block region, showing the three-nail pattern and the central screw. The holes were plugged with wooden dowels.

Endoscope view showing the central screw and the three plugged holes

FIGURE 4 3D rendering of the two merged volumes, showing the 4+3 nail patterns and the central screw in the original section of the top-block closer to the upper rib

FIGURE 5 Only the three-nail pattern and the central screw are present in the top-block section that was added later

FIGURE 6 3D rendering of the two merged CT volumes. This shows the original heel of the neck, with traces of the four-and three-nail patterns, plus the central screw.

FIGURE 7 3D rendering of the neck/body joint area. Later additions are represented in grey and the original features coloured brown. As well as the large ‘replacement cap’ on the top-block, additions to the neck (in the order installed) include: a small strip in the recess that originally hosted the top edge; an extension of the button; and a head-grain extension of the neck.

ALL CT SCANS FRANCESCO PIASENTINI @ TEC EUROLAB 

Rather than merely scrolling through CT slices, dedicated software also allows us to make a 3D rendering of a violin with various options. An example is shown in figures 4 and 5, where the two top-block sections mentioned above are recreated from inside the violin. The grey areas are regions where digital transparency was added in order to open the instrument to inspection virtually, inside its top-block.

All the images were made using VGStudio Max from Volume Graphics, a software dedicated to manipulation and analysis of CT volumes, thanks to the support of TEC Eurolab. This tool allowed the alignment and merge of the complete violin scan and the neck/ body area scan, with a resulting dataset that keeps the best from both scans. Small differences in grey values between the complete scan and the neck/body scan are evident in figures 4, 5 and 6, as well as the image on page 34.

The neck heel was also altered with respect to its original geometry. In its present condition, three wooden parts have been added in addition to the fingerboard: a small strip filling the slot formerly occupied by the belly’s upper edge; a wedge at the bottom of the heel; and finally, an extension to the headgrain of the neck heel to increase its length (figures 7 and 8). No nail hole traces are present in the added neck heel extension, showing the current neck setting involved removal and patching of the nail holes and their replacement with a metal screw. The virtual removal of these reveals what remains of the original neck.

FIGURE 8 3D rendering of the neck/body joint in the original part of the neck, showing (above) added parts and the 4+3 nailings in the central area, and (below) the added extension to the neck heel, with only the central screw.

FIGURE 9 A simple bidimensional scheme of the neck/body realignment procedure

After analysing the chronology of the neck and heel configuration, and digitally removing the non-original wood parts, we are able to realign the neck to the body of the instrument according to what was found to be the oldest configuration: the diamond-shaped four-nail pattern.

Using the remaining nail fragments and holes as guides, we virtually repositioned the neck so that each nail hole/fragment was properly aligned (figure 9). The neck and the body are three-dimensional objects, and so, too, the configuration of nail holes and fragments. It is interesting to note that the direction of the lateral nails converge at the centre of the neck. This orientation not only helped us to reduce the margin of error when performing the virtual alignment; this feature of the lateral nails are in fact what helped stabilise the neck attachment in the original instrument.

The convergence of the nails towards the centre of the neck is clear from figure 10. The advantage of this pattern is improved stability to the neck/body joint. The neck cannot be removed from the top-block unless the lateral nails are removed first.

Although the original location of the upper saddle is no longer available, we can estimate a neck length of 120.9mm (+-0.5mm) when referring to the actual fingerboard end. The neck inclination was measured as 2.5 degrees (+-0.2) between the actual neck/fingerboard plane and the gluing plane of the back. The backwards tilt of the neck results in a displacement of minus 6.4mm (+-0.5mm) at the end of the fingerboard by the top nut as compared to a plane projected from the top to rib gluing surface. Some margin must be allowed for variations in the thickness of the upper saddle.

Because the pegbox of the ‘Messiah’ retains its original varnish, we can measure with certainty the length of the neck including the upper saddle.

Another interesting feature of CT analysis is its use in visualising the geometry of the plates to the ribs’ gluing surfaces. The pristine condition of the ‘Messiah’ makes it an especially informative example. From figure 12 we can confirm that the back/rib gluing surfaces reasonably fall on a single plane, while the top/rib gluing surface appears to sit on a rather flat section, with uniform rib heights between the lower block to the four corner-blocks, before descending in a pronounced inclination through the upper bout to the top-block, around 2mm lower than the others. Whether these surfaces occur in two distinct planes or a single, large, sweeping but tapered arc calls for closer study. But CT analysis of the ‘Messiah’ confirms that the tapering of the ribs occurs in the upper bout of the top/rib surface, not the back, and that it involves some bending of the top plate as opposed to the flat taper used by many other makers.

FIGURE 10 Semi-transparent top and lateral view showing the alignment of the nail holes in the neck to those in the top-block. The three-dimensional requirements of this alignment results in a unique neck position that can be established with reasonable certainty.

FIGURE 11 A comparison between the actual neck set (transparent yellow) and the neck aligned according to the original four-nail pattern. The resulting neck length measured according to the actual fingerboard end is 120.9mm.

ALL CT SCANS FRANCESCO PIASENTINI @ TEC EUROLAB

FIGURE 12 The top gluing surface aligned behind a blue reference plane (when imagined in 3D) and the backward-tilted gluing surface of the neck to fingerboard, which begins on the plane at the rib to neck junction and descends to 6.4mm below the plane at the top nut.

Industrial CT is not just about on-screen diagnostics. In recent years, advances in technology have made it a powerful tool for recreating the complex surfaces of objects, including violins. In our study, the scans were ‘cleaned’ of non-original pieces added over the years, and the holes in the four-hole pattern were ‘emptied’ of metal and glue. The resulting pure surfaces were then exported as STL files and 3D-printed with an SLS printer in nylon (PA 6.6) and glass microspheres through the support of Andrea Scanavini and the team at Controllo Qualità Srl of Modena.

An operation that had previously been explored virtually on a computer screen can now be completed hands-on, using life-size and dimensionally accurate parts (figures 13 and 14). Fastening the ‘Messiah’ neck to its original position using its four geometrically aligned ‘nails’ opens a completely new experience of the ‘Messiah’ original set-up configuration, and lays the groundwork for further study. By collaborating with the MdV and its archive of Stradivari relics, we can continue to build-up this 3D model based on Stradivari’s original fingerboard, bridge, and tailpiece drawings, his original templates and fragments of original strings. Other experts with data drawn from access to original parts are invited to participate. The violin’s original fingerboard and set-up parts must surely still exist – somewhere!

FIGURE 13 STL files exported from CT data were 3D-printed in nylon and glass microspheres to create durable, dimensionally precise replicas of sections of the original ‘Messiah’

FIGURE 14 Dimensionally precise replicas of the body and the neck of the ‘Messiah’ are realigned in their original positions, using wooden splines shaped to fit the original nail holes.

Gregg Alf uses a strong light for the Arvedi Laboratory’s endoscope images on page 36

REPLICA PHOTOS COURTESY FRANCESCO PIASENTINI. ENDOSCOPE PHOTO ARVEDI LABORATORY OF NON-INVASIVE DIAGNOSTICS @ UNIVERSITY OF PAVIA

The virtual neck reset was made possible by means of advanced data manipulation. Advanced segmentation techniques were applied to remove later wood additions from the neck and top-block, much in the same manner as a restorer would have removed them at the bench. The neck was re-aligned according to what we have determined to be its original position, based on the plentiful and cross-referencing tracks from its original four-nail ‘diamond’ pattern. The same approach can be applied to the three-nail pattern, and will allow a comparison between each of the three different set-ups that the ‘Messiah’ has used over the past 300 years: the current set-up (with a screw), the intermediate (with three nails) and Stradivari’s original neck setting (with four). Further study of the screw geometry, its shape and possibly its metal content can provide information about its age and origin, and perhaps confirm a terminus post quem for its installation.

The authors hope that these results will open new perspectives in the use of CT technology for the study of rare stringed instruments. Other violins with similar features (nail traces, original necks) still exist, and may add useful information to what we know about the historical set-up of stringed instruments.

The authors wish to thank the Ashmolean Museum and senior curator Colin Harrison for sharing the ‘Messiah’ with us; the Museo del Violino and Fausto Cacciatori for including us in their programme of study; TEC Eurolab and Marco Moscatti for providing the tools (hardware and software) for the examination, and for help in fine-tuning the results; Academia Cremonensis, Cremona’s graduate school of violin making, and Giovanni Colonna for funding the CT study at TEC Eurolab