Is it impossible to hit perfect notes on fretless instruments?

A bit more on fretless and fretted instruments and beat counting.

Good points on beating. I have to think more about how I’ve actually done tuning (not for 20 years).

I: Cool refretted guitars, and compositions for consistent scales of non-standard series–overtone intervals, just, and for prime intervals, is at my friend’s site, http://users.rcn.com/dante.interport//justguitar.html. He has pages on the theory of the scales (traditional stuff), but also on the mechanics of how the guitars were built, and how ear-training to hear beats and tune the damn things.

II: What is interesting to me (my field, after performance, became academic musicology) is that before equal temperament, the whole topic of relative pitch concordance was really only theoretical. (Absolute pitches were not handled theoretically very much back then.)

In practice, different locales, room temps, or simply within a single band performance, created a mish-mash of tunings–you think Bach really gave a damn about discrete tuning systems (Werckmeister 2 or all that stuff, let alone waiting around for the flute to warm up and get back in pich) when in mid-concert he had to retune? Especially when he was playing with fixed-interval instruments? A nice jangling sound wall was/is pleasure enough.

Plus, IMO, the thread’s agreement of the use of vibrato as a compensating factor for fixed interval instruments is overblown (no pun intended).

The point of vibrato, or the once-fashionable stronger bowing or blowing and (what my teacher Ken Cooper called “banana notes”) is not to maintain a tuning, but a) not even sure if it existed at all in medieval performance practice, and b) was an “ornament” in later, pre-modern practices as style, never identified as intervalic.

However, for various reasons it’s not worth getting into here, the near-vibrato for minute changes in pitch in fretless instruments, which then became seriously futzing with pitch changes via finger positions, became “demanded” at discrete points in a composition. This was the accepted change from equal temperament beginning in the 19th century.

The reasons, basically, were changes in compositional style and in the overall construction of fretless string instruments. Players began to change the intervallic consistency between the leading tone and the tonic.

I apologize for bringing up these terms, which stem from an enormous field in music, but they seemed necessary. To go there would lead topic drift even farther than it is now…:slight_smile:

Something I was rather loose with terminology on was the difference between intonation and temperament. When it comes to guitars the two are rather interlinked. Becuse the guitar is a real life mechanical instrument, and yet has a rather rigid idealised design in many ways, the idea of intoning a guitar gets messy.

The nut of a guitar is usually treated as the location of a notional zeroth fret. (Some guitars actually have a nut and a zeroth fret. It is a rare thing however.) Like any real world instrument, you have a disparity with the location of the mechanical endpoint of the string and the physical location of the node of vibration (the witness point). The thicker the string the further in from the nut the witness point is located. Already there is a problem with a guitar, as the nut is a straight line, although the strings vary in thickness. Wound strings have a witness point closer to that of the substrate string. So the nut is already wrong.

When a guitar is played the strings are forced agaist the frets, this leads to a change in tension of the string, partly just to force the string out of a stright line, amd partly the pressure of the finger behind the fret on the string. So the change in tension comes partly in a definable form (from simple geometry of the guitar setup and the string relief) and partly from player’s style (the force a player tends to fret with.) Yet open strings have their frequency defined only by the location of the nut and bridge.

Depending upon your style of guitar the bridge can be of a wide variety of designs. A simple straight piece of material normal to the strings, a straight piece of material placed at a slight slant - and thus providing a first approximation to compensating for the different witness points. A two piece bridge for the wound and unwound strings, both at a slant. which provides better compensation for the witness points. And this goes all the way down to a guitar which provides individual length adjustments for each string. A luthier setting up even a classical guitar will usually file the bridge to change the effective length of each string - although the degree of adjustment is quite narrow.

Adjusting at the bridge is essentially used the to adjust for double the witness point difference, adjusting for the distance for both ends of the string at the bridge end. This is still an unsatifactory set of adjustments. The fret spacing is the same for every string and is derived from a notional string length, not the compensated string length. So the closer you play to the bridge, at least for some strings, the more out of tune.

Also, open strings are played without the additional pressure of fretting the string - and so may be slighty out of tune with the fretted notes. Many luthiers will move the nut very slightly closer to the frets to compensate for this.

So you basically end up with an instrument that is notionally designed to provide equal temperament, but actually is only an approximation. And sometimes not a very good approximation.

Then people start to fiddle.

The thing about a guitar is that it has a set of goemetric qualities that means it is played in a particular manner. One that is significantly different to a piano. For this discussion the most clear one is that the guitar presents the player with more than a single location for most notes. A guitar can span up to 4 octaves (for a 24 fret version) - which is 48 notes. But it has 6 strings, and that same 24 frets yields three times as many discrete frettable notes. On that same 24 fret guitar we can see than in the extreme, the low E is represented in only one place, whilst the high E is represented on every string. On average however, each note is found in about three places. And due to the earlier described mechanical issues, there is essentially no chance at all that those locations actually have the same frequency.

Now we can also note that the strings have discrete roles. The low E string produces either a bassline, or forms the root of many chords. The fifth string, when playing many chords provides the fifth. So, one thought experiement - tune the guitar so that rather than tune the open fifth string to a fourth higher then the sixth, tune it that subtle bit out so that when both are fretted a fifth apart, they are a perfect fifth apart. Power chords, here we come. :smiley: Once you start down this road you can go on forever. A change in bridge position can be used to cause a region of string to have the note intervals be compressed or stretched relative to ET. This may be deemed useful.

We can also note that the guitar favours different key signatures in different ways. One tends not to play chords in the higher fret positions - although they are technically available, the sound isn’t great. You lose the depth of sound when the string is fretted short, and the issues described earlier also push the harmonies out. And so it goes.

So - what does this mean? What one has is a deperately over constrained problem when it comes to getting a guitar in tune. And it is even worse when it comes to harmonies and getting chords to sound reasonable. But you can exploit the nature of the guitar. Offsetting the tuning of some strings away from the technically ideal equal temperment (at least as close as the device can get) allows you to approximate other desirable intonation attributes.

But, what it isn’t (and where the opening comment comes in); this isn’t a temperament. It is an intonation. There is no regular ratio of frequencies in an octave. The ratios from one octave to another will be different, and worse, the same note in the same octave, in different positions on the neck will have different frequencies. The only variables that are available to form this approximate intonation are string diameter, bridge position and string tension. An additional variable that a luthier can bring to the party is nut position. The Buzz Feiten system moves the but. There are other nut variants. Earvanva do a staggered nut that is intended to address the different witness point issue.

Also what there isn’t, is much of a recognised technical understanding of this. Before Buzz Feiten patented his ideas, luthiers had their own personel set of tricks, and experienced ones played with the intonation to suit a player, often simply based upon empirical experience. The Feiten system is still probably the only well codified system. A guitar setup by an experienced luthier can often involve a subset of these ideas. One guy know will set up the intonation to be more consonant at different parts of the neck to suit different styles. Guitar players will also simply find that a particular way of tuning the guitar works better for them than others. Some like to tune with harmonics, some will just use a tuner, others just run down the fretted forths. Some care a lot more than others. Similarly, setting the bridge position. The classic way it to match the frequency when fretted at the 12th fret with the frequency of the second harmonic. But some people find that other rules seems to work better for them. Again, empirical knowledge. But you end up with an intonation that suits the player, and takes the already imperfect approximation the guitar has to ET, to some other intonation, that has some aspects of ET, but also probably some significant variation.

In truth, for 99% of guitar players, no one cares. The guitar is an imperfect approximation to ET, and players learn to avoid those parts of the guitar that sound poor. But the other 1% that do care seem to evolve a set of personal changes to the standard setup that serves them well. What is missing from the world is much of a codification of this.

Thank you! Most informative.
.

I’m not following the explanation and how it is directed at the question. The mathematical equal temperament frequencies for the octave above and below A-440 are as follows. Looks like stpauler has the right frequency for C when rounded to two places:

220.000 A
233.082
246.942 B
261.626 C
277.183
293.665 D
311.127
329.628 E
349.228 F
369.994
391.995 G
415.305
440.000 A
466.164
493.883 B
523.251 C
554.365
587.330 D
622.254
659.255 E
698.456 F
739.989
783.991 G
830.609
880.000 A

So how would a middle C actually be tuned?

I am not familiar with the concept of the witness point. I had been under the impression that a vibration string was forced by its circumstances to have nodes exactly at its endpoints. Can you explain further or provide a reference for further research by the student? :slight_smile:

I used to tune using harmonics and was never happy with the results. I use a tuner with great results, but I also get very good results by finding an A on every string and tuning it against the open A string. After having read your treatise and learning about witness point reconciliation, however, I am now not so convinced that is best.

I always thought that was the textbook method

but I guess other people bought different textbooks :slight_smile:

A Tube Screamer covers a multitude of tuning sins :smiley: (Actually I say that in jest because I have found that in some situations, particular playing a double stop, a distortion device can magnify tuning flaws.)

Thanks for a thoughtful and informative post!

The frequencies listed above are indeed correct, and if you were to tune an organ, harpsichord, or guitar precisely to those pitches the result would be a perfect tuning.

But for a piano, it’s only a good starting point; a piano tuned precisely to these frequencies will not be in very fine tune. Some notes might in fact conform to the list, but probably not most of them, and definitely not all of them.

As I tried to explain earlier, inharmonicity is the reason. The same effect that forces us to stretch-tune piano octaves also forces adjustments to smaller intervals. See my earlier post explaining why, for example, A-220Hz will be too sharp, and A-880Hz too flat, for a 440Hz middle A.

“How would middle C actually be tuned?”

If you mean other than 261.626Hz? There is no one answer, and that’s my point. It depends on the specific piano being tuned. The amount of inharmonicity varies from one area of the keyboard to another, and from one piano to another, unless the pianos are identical in design and condition.

This is something I often found myself attempting to explain to various clients, and my experience was that many, or even most people, can find it quite difficult to wrap their heads around the concept of inharmonicity in pianos.

I also found that I’m not very good at articulating it, so if my earlier post is still inadequate, I suggest taking a look at the link I provided.

.

CookingWithGas, an ideal string tuned to A440 will have a harmonic at 880.000 Hz. A real piano string will have a harmonic somewhat sharp, say 890 Hz. If you tune the A880 strings to 880.000 Hz, it will be out of tune relative to the harmonic of the A440 string. So you tune it a bit sharp to 890 Hz, and it’s in tune with the harmonic of the A440 string.

Similarly, all the strings other than A440 would need to be tuned a little sharp or flat, to make the harmonics of lower strings in tune with higher strings.

Well, that’s simple enough. Thanks.

Here’s a reference(one that also happens to be on my shelf) (see first paragraph on the displayed page) that equates the mechanical endpoint of the string with the witness point. Are there sources that address this differently?

Nonsense. Of course they are. Do you know what equal temperament is?

I do know. I miswrote - they are not intoned to equal temperament. In particular stretch tuning prevents this. But the divisions within an octave are close to equal, and thus very close to the 12th root of 2, and so we can say that pianos seek to approximate ET.

As I wrote above, whilst the simple design of a guitar would suggest ET, the reality is that most guitars are at best a poor approximation to ET, and many players and luthiers, either by design, or by personally evolved set of tweaks, create an intonation that is a piecewise amalgam of bits of ET with different offsets and different errors.

This statement is ridiculously wrong.

When a piano is properly tuned in equal temperament, the divisions within the octaves ARE equal. The RATIO of any two adjacent notes is the same throughout the scale. Inharmonicity affects the size of some intervals, and stretch tuning accomodates the changes, but in it most definitely does not prevent a piano from being in ET.

.

I don’t think you realize it but you’re using the phrase ET in a way that’s very hyper academic and too precise. It’s a definition that ignores everyday usage of the phrase “equal temperment” and your explanation (while correct in a ultra scientific way) does not allow for any musical instruments to ever be flawless ET.

This scientific definition is causing the confusion like the statements below.

To give a similar example, let’s say I ask you to look at the THE STRAIGHT DOPE LOGO at the top of this web page. Do you see a straight line? If you say yes, I say you’re wrong because there ARE NO TRULY STRAIGHT LINES that exists anywhere in the universe. A true straight line is a mathematical construct. If you were to zoom into the computer monitor and examine the row pixels that supposedly make up that straight line, you’ll see they are jagged and not perfectly straight at all. You might concede that yes, it’s a mathematically accurate definition of a “straight line” but it doesn’t necessarily help the conversation. Therefore, insisting that the “almost straight” row of pixels is “seeking to approximate a straight line” is going to make people scratch their heads.

TreacherousCretin and CookingWithGas are using the phrase “equal temperment” the way most people use it. It’s a tuning that’s based off of the 12th root of 2 even if some of the notes don’t exactly line up with the mathematical calculations.

So for most folks, piano “stretch tuning” or guitar “Buzz Feiten tuning” are still “based” on equal temperment. It would actually add to the confusion to say they are not equal temperment. For most folks, examples of a tuning that’s not equal temperment would be 11th-root-of-2 or 31st-root-of-2 or “just intonation based on whole number ratios.”

Hmm, I think I’m getting confused. And as above maybe I’m being simply too prescriptive (heck, I am an academic.)

A though experiment. If you tune a short piano, and apply approriate stretching, and a very long piano, also with approriate stretching, they won’t be in tune with one another. So, are both ET? If you are asked to tune two pianos, of diferent sizes, to be played as a duet, what do you tune them to? One assume that in general you would want to avoid such a situation.

I’m not sure what the answer is, but I’m trying to get some idea of the scope of what happens.

So, ignoring the first rule of holes, I thought I would dig myself in a bit deeper.

Yup. Exactly true. But if you look back at the OP’s question, that was basically the point of the discussion. The variance from infinite precision. In that vein I’m being ultra precise. Mostly because I have always found this subject rather fascinating.

So, I’m being probably annoyingly pedantic, but also learning some interesting stuff. So, stretch tuning. I had to look around for an available on-line reference, this version of a Peterson tuner manual has stretch tables for a range of pianos. pianosupply.com - This website is for sale! - pianosupply Resources and Information. The point I was making rather badly, was that across a piano, the individual octaves have different ratios. On large pianos quite a wide range of notes are not streched, indeed we see from C2 to C6 on the concert grand, essentially no stretch at all, but stretch that widens octaves either side. So there are essentially two sorts of octave. I have a suspicion, and maybe TreacherousCretin would be able to confirm that the breakpoint in inharmonicity that is being compensated for in the bass corresponds to a change in string construction. Looking at a more common, shorter frame piano the stretch gets much larger and affects more notes. Indeed it quickly gets to the point where there are probably 3 different stretch coefficients if one wanted to do a reasonable piecewise linear approximation to the curve. So it is clear that for many pianos one could find an octave in which the ratios between the notes actually changed from one end to the other. Again, I am suspicious that some of the these breakpoints correspond to changes in string construction.

Vibrating string physics is interesting. The usual description we see, and the classical (D’Alembert) derivation of the vibration of a string assumes a zero diameter string, asssumes the string stays the same length (despite being stretched in order to vibrate), and assumes no change in other physical parameters. Once you add a non-zero diameter, you are faced with a string that has stiffness. At the ends of the string you have an interesting boundary condition. The string is held solidly at the end - either the nut of the bridge, and is straight. The vibrating string must terminate in a straight peice of string. The interesting upshot is that, despite the ususal drawings of a vibrating string showing a convex curve, the actual shape at the absolute end of the string is concave. Thus the string behaves as if it is shorter than the distance between the ends. This stiffness also affects the vibration along the length of the string. The other interesting property usually ignored is dispersion. The vibration of the string propagates along the string at the speed of sound in that material. But this speed isn’t constant with frequency. All this conspires to move the harmonics to frequencies that are no longer harmonic - and they become partials.

Electric guitars are interesting, in that they are often played though a distorting amplifer, and other devices (i.e. a Tube Screamer as mentioned earlier.) The really rotten problem that this results in is that the distortion chain creates harmonically correct overtones, whilst the physics of the strings adds harmonically wrong partials. Great. One or the other might be OK, but the two together gets you a set of very closely spaced frequencies, and some real dissonance. Also, a guitar that has been intoned to a close approximation of ET has the harmonics from other strings in the chord in the wrong place. So it can end up a mess. This is why some guitarists will use a “power fifths” tuning. Again, to reiterate the earlier discussion. The guitar can be tuned so that the gaps between strings is no longer an ET 4th. Rather the strings are tuned so that fretted power chords (1sts and 5ths) are in tune.
This is a hybrid temperament. One could quite reasonably say that for melodic purposes the guitar remains ET, but for harmonic purposes it is no longer ET. Chords played on a guitar so tuned will not have the same tonality as chords constructed from notes in the ET scale. And that is IMHO a pretty significant issue. It is this manner in which the individual chord tones can be tweaked - by virtue of the layout of strings - that allows the guitar to play harmonically from a different tempering than ET. Melodically, if the melody stays on one string, it remains ET.

This is true and I’ll bet many guitarists don’t even understand why they do this. Rock players crunch on power chords, which sound great through distortion devices (need I mention “Smoke On The Water”?). But take that same guitar and distortion device and play any chord on the top three strings, or even just a major third on the second and third strings, and it will sound like someone stuck a pin in a monkey’s ass.