note: non promotional article
A HARPSICHORD BUYER'S GUIDE
Part I (to be continued)
A well made harpsichord should last, and give satisfaction, for a lifetime, just as did the best historical instruments. Unfortunately I often come across modern instruments that fail in these respects. This section is created to help you avoiding throwing away your hard earned money when choosing a harpsichord. When a violinist buys a violin he will look carefully at the quality of the workmanship as well as listening to the sound quality. Before buying a house it is normal to get a survey to check the soundness of its structure. You should do the same when buying a harpsichord. But what to look for? Here is my answer, based on 25 years of experience.
The first thing I would advise if you are new to harpsichords is not to decide on a builder until you have seen and tried the instruments of several builders. Of course you must like the sound, not just straight away, but also after half an hour of playing on the instrument, in which you have listened to each register in the bass, tenor, middle and treble regions. Here are some questions to ask yourself (and the builder!):
1) Is it based on a historical instrument, and if so, how close is it to the original?
Of course a builder can invent his own instrument, and it could be very good, but unfortunately most of the problems I have encountered in harpsichords have come up where the builder has departed from an original. The builder has put in an extra 8' register in a Ruckers? (the originals had one 8' and one 4', even the doubles). The result is often that the extra tension causes the bentside to bend inwards, pushing the soundboard into a bulge that can touch the 4' strings in damp weather (I have seen this several times). I once saw a 'copy' of a 17th century French harpsichord made by a famous maker. I wondered why he had put in double bridge pins in the 8' bridge (the original does not have any). The reason was that the builder had changed the soundboard ribbing (one could feel that the soundboard was more flexible than it should have been), the bridge was tipping over and the builder had to correct it by putting in the double bridge pins.
So wherever possible I would recommend you should go and visit the original harpsichord (see the original instrument map to find the original instruments you can visit). Look carefully at the case . How are the case sides joined together (dovetails will be visible through the decoration)? The curve of the bentside. The keyboard, the curves of the bridges, the distribution of the strings.
The sound quality of a harpsichord depends largely on
the quality of the materials used.
No.1 rule: avoid plywood at all costs! This is mainly because of the horrible acoustic properties of plywood, the layers of glue deaden the sound. The wood in commercial plywood is not chosen for its acoustic properties, very often African hardwoods are used, which combine weight with a sort of deadness (acoustically speaking). If in a painted harpsichord, on lifting the lid to open it, you are surprised by its weight, then probably plywood is the culprit. Some plywoods also suffer from "creep", that is when under tension the layers of wood move slightly over each other allowing the wood to bend. How do you recognize plywood? Modern plywood is made by spinning a log of wood and peeling off a layer with a long knife. So the wood does not have the characteristic grain pattern of ordinary wood. If you look at the baseboard of the harpsichord from underneath you will see the difference between a baseboard glued up out of planks, and the almost grainless look of a sheet of plywood. If the builder has used plywood in the baseboard, it is possible he has used it elsewhere. But even the quality of the baseboard is important for the sound. The harpsichord case is a resonating box: one would not buy a cello with a plywood back. Try tapping the bottom of the instrument. There should be a resonant ring to the sound. The same considerations apply to the outer case of an Italian harpsichord.
Northern European soundboards were made of European spruce, to use an approximate term. As these species are not indigenous to Britain (it's a strange thing that they never arrived in Britain after the ice age), there is no exact term for them in the English language. It is best therefore to use the Latin names: there are two species, Picea abies (sometimes known as Picea excelsa), which grows in the Alps and in Scandinavia, and Abies Alba, which grows at somewhat lower altitudes. The other European languages have exact names for these trees: Picea Abies is 'fichte' in German and 'abete rosso' in Italian, abies alba is 'tanne' in German and 'abete bianco' in Italian. The boards used for soundboards should be radially cut, giving a pattern of straight lines to the wood grain instead of curves. The colour is whitish when cut, ageing over years to a honey colour. American or Canadian wood (which sometimes has a pinkish or yellowish colour, but not always) should be avoided.
Italian builders used cypress, but from the end of the 17th century they sometimes began to use spruce as well. In my experience cypress has a less resonant, but perhaps sweeter sound than spruce.
3)The quality of the workmanship
Look carefully at the soundboard. Because the wood is radially cut, and the centre of the log cannot be used, the planks are quite narrow, and the soundboard is normally made by gluing up ten or more planks. But the joins should be nearly invisible, you should not be able to see any glue line between them, and certainly not any gaps! Do not buy an instrument with a split or gap in the soundboard. The treble of the soundboard needs special attention when preparing, to avoid splits forming in time. Normally in furniture making, panels of wood were left free to expand and contract with changes of humidity by being slotted into a frame without glueing. The frame was designed to take the load, and the panel to fill the space. However, for a harpsichord, this form of construction is not possible, the harpsichord case is a sound box and the treble portion of the soundboard is the most dangerous area, as it is glued to the case edge and the belly rail (or header rail). Thus it is constrained by these two parts and in dry weather will tend to shrink compared with them. If the builder has not taken special precautions before gluing in the soundboard, it is certain to split after a few years, unless kept at absolutely constant humidity. The most likely place for splitting is just to the right of the end of the bridge. If the instrument is already a few years old and not cracked (if in doubt, the soundboard should be honey-coloured, not whitish) that is a good sign. If you are buying the instrument from the maker, ask him about this; how did he prepare the soundboard before gluing it in?
4)The structure: Is the case strong enough?
This is a most important factor in choosing an
instrument. The case of a harpsichord is under a continuous tension
and the case/frame should be well able to cope with this. Of course
wood is wood, and a certain amount of distortion is inevitable.
Straight lines become slightly curved over time, but this should not
be more than a few millimetres. One good test is to look at the
instrument from the side, lining up with your eye the top edges of
the cheekpiece and spine. You will probably see that the top corner
of the cheekpiece/bentside join has been pulled a bit down. This is
because the strings being near the top of the case, they tend to
twist the wrestblock clockwise (as seen from the cheekpiece) and with
it the cheekpiece. This twisting should not be more than 2 or 3
millimetres on a well built harpsichord. On my 1982 Taskin, this
twisting is barely visible, not more than 1 mm.
Another point to look out for is the distortion of the bentsidecaused by the tension of the strings. This can be seen by putting a ruler held vertically against the outside of the bentside about halfway along its length. This distortion should not be more than a few millimetres. This bending of the bentside has the effect of squeezing the soundboard, distorting it and tipping the bridge over. In extreme cases this can lead to the 4' strings touching the soundboard. It was to avoid this that builders normally put ribs across the soundboard before the 18th century. The exception was the Ruckers family, who introduced a design without ribs between the case side and the 4' hitchpin rail. The case must be strong enough to take the tension of the strings unaided by the soundboard. If however we load a Ruckers case with an extra 8' choir of strings, we are heading for trouble as mentioned above. If a builder thinks he is improving a French 17th century design by putting in Ruckers style ribbing, he is also heading for trouble. In the 18th century, French and German builders also used Ruckers style ribbing, but the cases became much stronger with thicker sides, and stronger frames. Italian instruments continued to use cross ribbing except where the soundboard grain is diagonal (such as the Grimaldi) where the soundboard itself can aid much more in resisting the pressure of the bentside.
Flemish, French and some German harpsichords (such as the Fleischer) are built by first assembling the case sides with the wrestplank. You should check that the wrestplank is neatly jointed to the case sides. Look under the wrestplank by taking out the name batten, and perhaps the upper manual in a double. In 18th century harpsichords, the case sides are jointed together with dovetail joints. This is a marvellous joint, and well worth the trouble to make. You should be able to see the outlines of the dovetails even if the case is painted, because of the slight movement of the wood with changes in humidity.
Italian harpsichords, and some German harpsichords are assembled in the reverse order. The frame, including the wrestblock, is assembled on the baseboard. The sides are glued to this frame. The wrestblock is fixed onto support blocks glued to the baseboard. If you take out the frontboard you will be able to check that these joints are sound. There should be no gap between the wrestblock and its support blocks. The tension of the strings tends to pull the wrestblock up into a curve, and you should check that this is not more than very slight.
It is clear that the old builders took care to ensure that the string scalings, the lengths of the strings, were those that gave the best sound for their instruments. The builder of your instrument should have done the same. I would strongly recommend you to take a tape measure, possibly also a pocket calculator, with you when you go to see an instrument, and actually measure the speaking lengths of some of the strings, that is the distance between the nut pin and the bridge pin. However much you feel you trust the builder, before spending all that money why not check for yourself? We are going to examine the four notes c, c' (middle c) c'' and c''', that is the string of the longer 8' register. To make the measurements you should take off the jackrail and depress the key to find the jacks and strings for that note. Here are a few examples of original instruments (in no particular order) that represent harpsichord building over a wide area of Western Europe:
Lengths of the longer 8' strings of some well known historical harpsichords
The main reason I have put in this table is to show that for all these builders, from c'' upwards they were careful to follow a very nearly exact scaling (that is halving the string length when going up an octave). the Dulcken, Mietke, Cristofori and Zell are exact within the limits of measurement, the Delin Fleischer Hemsch and Ruckers are within 3 millimetres, or 1/8 of an inch. These makers knew the importance of correct scaling to obtain that bright clear treble sound. The shorter string lengths of the Mietke and Cristofori are explained by the use of brass strings instead of iron. Apart from that, variations in string length can also be explained by differences in the pitch used in different places and times. For instance it is clear that the Dulcken and Hemsch instruments were made to play at a low pitch (lower than 415) , the Ruckers and Fleischer would work well at around 415 whereas the Zell and Delin would work well at 440.
I have not used data from earlier instruments as nearly all the 16th century instruments, and many from the 17th century have been altered during rebuildings and restorations, so that their original scalings are doubtful.
If you have your pocket calculator with you, you might like to check the length of the string of the top note of the instrument you are examining. For instance, f''' should be three quarters of the length of c''' ; for e''' multiply the length of c''' by 0.79 . The lengths you measure should not be shorter (give or take a couple of millimetres) than these calculated lengths. However it seems that some makers (such as Taskin) made use of an effect called tensile pickup, whereby drawing iron strings causes the material to harden, so that the very fine top strings can be a bit longer than the calculated lengths. This effect does not apply to brass strings.
Of course slacker strings break less often, but I would much rather have to occasionally change a string (its no big deal) and have a satisfying sound, rather than slack strings with a dull sound.
Going down an octave from c'' to c' the string length should double, but in practice most makers began to foreshorten the strings slightly. For the Dulcken, Mietke, Zell, Delin , Hemsch and Ruckers the string length of c' is between 95% and 97% of what it would be for an exact doubling whereas Cristofori and Fleischer doubled their string lengths, in the case of Cristofori with great precision.
Going down another octave to c, there is some more foreshortening in the string lengths. The c lengths of the Ruckers and the Dulcken are only 79% of a theoretical quadrupling compared to c''. The other instruments are in the range of 80%-90% except for the Mietke and the Cristofori (95% and 99%). The foreshortening is compensated with heavier strings.
Going further down, string lengths depend on the length of the instrument and whether there is a changeover from iron to brass.
As a conclusion, for a 415 instrument strung in iron,
the c'' length should be between 35cm and 37cm (give or take a few
millimetres). For brass it should be between 27cm and 29cm. It is no
use copying an instrument made to play at a lower pitch, such as the
Dulcken, without redesigning it for the intended pitch. At 440 these
numbers should be reduced by 2cm.
Above c'' there should be no, or at most very little foreshortening, not more than 3mm for c'''. Going down from c'', the foreshortening should gradually increase as discussed above. It will be very useful for you to actually check these measurements on an instrument you are trying, and consider their influence on sound quality. Perhaps if the sound of the top notes is a little dull, could it be due to the scaling?
If you have any doubts on the scaling of an instrument, you might care to drop me an email with the measurements you have made. If possible I can check the measurements against those of the original harpsichord.
I have not discussed here the diameters of the strings for three reasons:
First, the original diameters are often not known. Sometimes gauge numbers are stamped on the wrestblock, but often the interpretation of these numbers is controversial.
Second, strings can be changed if it is thought that a different gauge might improve the sound. Changing the speaking length of a string is altogether a different matter!
Thirdly, for a given material, the pitch at which a string will break does not depend on its diameter, only on the sounding length (with the possible exeption of fine iron gauges subject to tensile pickup), so that the optimum length for a good sound does not depend on the string diameter. Conversely, knowing the length of a string on an old instrument does not enable us to determine the string diameters used, only the string material if we have an idea of the pitch at which the instrument might have played.
In a future article I will talk about the harpsichord action.
Please send me with your comments about this article.
Tony Chinnery Via Padule, 93 Vicchio(FI) 50039 Italy Tel:39-0558407817
new email address: