ASA/CAA '05 Meeting, Vancouver, BC


[ Lay Language Paper Index | Press Room ]


Harp Design and Construction

Chris Waltham waltham@physics.ubc.ca
Department of Physics & Astronomy
University of British Columbia
Vancouver BC, Canada V6T 1Z1

Popular version of paper 5aMU8
Presented Friday morning, May 20th, 2005
ASA/CAA '05 Meeting, Vancouver, BC

Construction

The harp is triangular in shape. The easiest part of the triangle to make is the post (fig. 1), as it plays little part in the sound production and can therefore be over-engineered; the compressional force is in any case mostly axial. The neck has to withstand the total tension of all the strings, and also a large torque, as all the strings are mounted on one side. The curvature does not help, and as much of the characteristic grace and elegance of a harp derives from the neck shape, it cannot be overbuilt. Very strong many-layered plywood seems to be the best material, covered with veneer for beauty's sake. The soundboard has to be both thin (note how floppy it is), and also to withstand all the string tension (6000N in this case - more than half a ton). Sitka spruce is the material of choice, for its acoustical properties and anisotropic strength. The grain runs horizontally, and thin veneer with vertical grain  is applied to prevent cracking (although it degrades the sound slightly). The soundbox is a light, hollow shell, with holes at the back for improved sound radiation and access to the strings. The whole structure bends alarmingly under the string tension (and occasionally explodes), so the veneer has to be very well bonded.

Figure 1: Construction Sequence
post
capital
neck
soundboard soundbox sb_centre
Post capital and base on lathe
(birch)
Mortising the post capital
Neck
(baltic birch ply, birch/maple veneer)
Soundboard
(sitka spruce)
Soundbox (birch/maple veneer on "rubber" ply) String support
(oak and nylon)

The Finished Product

In the following panel I have placed my own celtic and lever harps in the context of  harp evolution (fig. 2). The harp has been basically triangular-shaped for about 1000 years. The gothic harps had small, thick soundboxes and soundboards carved out of two solid pieces of wood, and needed "brays" to buzz against the strings and increase the sound output. Larger, more efficient soundboards came with the celtic harps. In the renaissance, soundboards were made ever thinner, especially in Spain. Chromatic tuning was achieved by having two or three rows of strings, making the harps very difficult to play, and multiplying the total force on the soundboard. Sharpening levers to raise the pitch of the strings by a semitone went some way to solving this problem. The ultimate "modern" concert harp was developed by Erard in Paris and London around 1800; these had pedals attached to a complex mechanism which could raise the pitch of the strings by one or two semitones. 

Figure 2: Evolution of Harps
Small soundbox,
thick soundboard
Larger soundbox
2,3 string rows:
chromatic
Levers for semitone sharpening,
thin soundboard

Double-action pedals
gothic_wien
harpI arpa a tre file HarpIII
erard wien
Gothic Harp (C15),Hofburg Museum,
Vienna
Homemade Celtic harp (18 strings) Arpa a tre file (1625),
Museo Civico,
Bologna
Homemade lever harp (36 strings),
copy of George Morley model (London, 1820)
Erard Harp c.1800 (modern concert harp), Hofburg Museum, Vienna

Strings

The string material is determined by harmonicity (the overtones should sound pleasant) and "feel" (fig. 3). Harmonicity requires that the string be strong, heavy enough, but not stiff. "Feel" is how hard the player has to pull to move the string centre a given amount before release; it should be large enough that the strings do not touch, and not vary too much from string to string. Nylon or gut would probably suffice for all strings if the lower strings followed the curve set by the upper strings and became very much longer than they actually are. Gut is mechanically similar to nylon but has a warmer tone for the mid-register. Practical reasons make the harp neck a double curve (an ogive, so it isn't too tall) and so much heavier strings are needed for the low register. Nylon/gut strings would have to be very thick here, very inharmonic, and the feel would be so small that the strings would interfere with each other. Steel wrapped with helical copper wire is used, although there is an awkward change in feel at the break.         

Figure 3: String data measured on a Salvi Aurora Concert Harp
salvi_length
salvi_stress
Basic harp shape and position of string types
Maximum stress (and tension) occurs in the steel strings
salvi_feel
salvi_inharmonicity
The highest steel and nylon strings feel the tightest to the player
The steel keeps the inharmonicity low for the lowest strings


The Characteristic Sound of a Harp

The ingredients which distinguish the sound of a harp from any other plucked instrument, say the guitar, are as follows:
  • The string is plucked in the middle, which reduces even harmonics. In some music, the harp is sometimes plucked close to the soundboard in order to imitate a guitar.
  • There is very strong coupling, via to soundboard, between any two strings with overlapping overtones. In the case of a large harp, many strings vibrate when only one is plucked; see figs. 4 and 5.
  • The long strings vibrate in collapsing and expanding ellipses, which gives many strings a long pulsed tail to the sound (fig. 6).
  • The large number of strings allows for glissandi, a feature most people immediately associate with the harp.

Figure 4: Sonograms produced by the A2 string on my lever harp and a guitar: which one is which?
The horizontal axis is frequency (pitch) with the left-most spike being the fundamental at 110 Hz; the other spikes are overtones.
The vertical axis (scrolling dowards) is time. The bright horizontal line is the inital pluck, which excites other string via soundboard resonances ("blobs" in the line). Faint vertical smudge around 60 Hz is caused by external noise. Click on images to hear sound.
A2
A2

Not so easy is it? Even when you hear it - without visual clues - it is not completely obvious. Look at the spaces between the overtones; the harp has many faint spikes caused by other strings starting to resonate. There is very string one at 262 Hz from the C4 string (an octave and a minor third away - a factor of  about 7/3).

Figure 5: Demonstration of String-Soundboard Coupling
Figure 6: Typical Motion of Centre of String
- snapshots every 1/2 second after plucking
harpII
D4
Click on image to see movie shot with 65Hz strobe. When string C2 (leftmost red string), note how C3 and C4 (also red) respond.
The large polygon at time zero is responsible for the initial attack, followed by the collapsing and expanding ellipses which produce the sustained, pulsing sound.
Acknowledgments
  • String motion data taken by Gary Chan, as part of his undergraduate thesis work; thanks to Andrzej Kotlicki for the position sensors.
  • Thanks to the UBC Music Department for the loan of the Salvi Aurora.
  • The sonogram software was obtained from www.baudline.com.
  • Harp plans from Robinson Harps, California.


[ Lay Language Paper Index | Press Room ]