Your guitar is intonated at the 12th fret. Every string checks out. You play an open E chord and the G sharp on the third string sounds wrong against the open B, and it has sounded wrong on every guitar you have owned.
The check is not lying. It is answering a different question than the one you have. A 12th-fret intonation adjustment calibrates the ratio between one fretted note and the open string, and it is structurally incapable of seeing an error that changes as you move along the neck. The error you are hearing changes as you move along the neck. It lives at the other end of the string.
The Short Answer
| Symptom | Cause | Fix |
|---|---|---|
| Fret 1-3 sharp, 12th fret perfect | Nut slots cut too high | Cut slots to correct depth. Biggest available win. |
| Still sharp after slots are correct, by a few cents | Nut position — inherent compensation error | Compensated nut, or accept it |
| Sharp everywhere, worse as you go up | Saddle position | Standard 12th-fret intonation |
| One string sharp, others fine, inconsistent readings | Dead or false string | Replace before measuring anything |
Do these in order. Every one of them is invisible to the check above it.
Why the Saddle Cannot Reach It
Moving the saddle changes the speaking length of every fretted note by the same physical distance. What that distance is worth, in pitch, depends on how long the note is to begin with.
On a 25.5-inch scale, the 12th fret sits at 12.75 inches from the saddle and the first fret sits at about 24.07 inches. Move the saddle back by ten thousandths of an inch and retune, and you have flattened the 12th fret by about 1.4 cents and the first fret by about 0.7. The correction is roughly twice as strong at the 12th fret as at the first — which is the exact opposite of the error curve you are trying to cancel.
So the bridge end can only make things worse in a specific, calculable way. To pull the first fret down by a cent you would push the twelfth down by two, and you would be trading a problem you can hear in open chords for a problem you can hear in every solo. That is not a tuning compromise. That is a worse guitar.
This is the same reasoning that makes the saddle-by-saddle intonation procedure a procedure about the 12th fret and nothing else. The saddle owns one variable. The nut owns the other one.
Where the Sharpness Comes From
Fretting a string does two things: it shortens the vibrating length, which is what the fret positions are calculated for, and it stretches the string, which nothing is calculated for. The stretch raises tension, and raised tension raises pitch. Every fretted note on every guitar is slightly sharp for this reason.
The amount is not constant. It depends on the geometry of the bend, and near the nut the geometry is severe. Press the string at the first fret and the string has to leave the nut, drop to the fingerboard across roughly an inch and a half, and continue to the bridge. That is a sharp angle over a short span. Press at the twelfth and the string is bending gently across two spans of about twelve inches each. The same downward deflection buys a much larger length change near the nut than it does in the middle of the neck.
That gives the error a shape: worst at the first fret, decaying steadily as you move up, and small enough by the seventh or eighth fret that nobody argues about it. It also explains why the complaint is always about open chords specifically. In first position you are sounding your sharpest fretted notes simultaneously with open strings that are exactly in tune. There is no more exposed arrangement available.
The Test That Separates the Two Causes
There are two separate things at the nut end and they need different responses. The slot depth is a setup error. The nut position is a design compromise. Almost everyone has some of the first and all guitars have the second.
Step 1. Replace the strings. A string that has gone false produces inconsistent readings at every fret and will waste the next twenty minutes. Our piece on when strings go false has the consistency check if you are not sure.
Step 2. Set the slot depth correctly. Use the feeler gauge and fret-rock method from the nut slot depth guide. This is the entire ballgame for most guitars. A slot left a few thousandths high forces extra deflection at exactly the fret where extra deflection costs the most, and a badly cut nut can push the first fret sharp by tens of cents — an order of magnitude more than the compensation error underneath it.
Step 3. Remeasure. Tune the open string with an accurate tuner, ideally a strobe. Fret the first fret with normal pressure and read the cents. Do all six. Write them down; this is the number that matters and it is yours, not a published average.
Step 4. Take the nut out of the circuit. Clamp the lightest capo you own directly behind the first fret. Retune to that pitch, then check the second fret. If the error largely disappears, the nut end was responsible for it and your frets and saddle are fine. A capo has its own height and its own stretch, so treat this as directional rather than absolute — it tells you which end of the string to look at, not how many cents to correct.
What is left after step 2 is the inherent compensation error, and on a well-set-up instrument it is usually a few cents. Most people can distinguish a pitch difference of somewhere around five to six cents. Hold those two numbers next to each other before you spend money.
What the Compensated Nut Systems Actually Move
A shelf nut moves the string's take-off point closer to the bridge. The frets stay where they are, so the open string gets shorter and you retune it down, and every fretted note on that string comes out flat by the same ratio.
Run the arithmetic on a typical offset. Shorten a 25.5-inch string by 0.020 inches and you have flattened every fretted note on it by about 1.4 cents. Uniformly. At the first fret and at the fifteenth.
Two problems fall out of that immediately.
The correction is flat and the error is a curve. You want a lot at the first fret and almost none at the twelfth. You get the same everywhere. Over-correct enough to fix first position and you have pushed the upper neck flat.
The correction is uniform across strings and the requirement is not. A stiffer string changes pitch more for the same deflection, so the strings do not need the same offset. One luthier's measurements put the high E at roughly half the sensitivity of the G. A single shelf applied to all six over-corrects some and under-corrects others, and published per-string figures from different sources disagree with each other by a factor of several — which is a useful thing to know before treating any of them as a specification.
This is why the Buzz Feiten system is three interventions and not one: a nut offset, altered saddle positions, and a tempered tuning chart the player has to use. The third exists because the first two do not fully agree. It is a real engineering compromise and it is presented as one by the people who install it, who will tell you that the D, A and high E can end up flat at the low frets as the price of getting the rest closer.
Earvana takes the other route — a two-piece nut with per-string offsets, so the correction is mechanical rather than procedural. That addresses the per-string half of the problem and leaves the per-fret half where it was. Its specific offset values are disputed by luthiers who have measured their own, which is another reason to test your own guitar rather than trusting a chart.
The String That Fails Worst Is the One You Would Guess
I expected the low E to be the outlier, because it is the thickest string and it takes the most force to fret. It is not. The plain G is.
The property that matters is not diameter, it is core stiffness relative to tension. A wound string's core is much thinner than the string looks, and the winding contributes mass without contributing much stiffness. The plain G has the largest solid core of any string in a standard set, at roughly the same tension as its neighbors, which makes it the string whose pitch moves most for a given deflection.
Which means the plain G is the worst offender at the nut end for exactly the same reason it is the outlier at the saddle end — the reason we laid out in the wound G versus plain G piece. Same property, opposite ends of the string, and the two failures are usually described as unrelated. They are the same failure. A player who switches to a wound third for the saddle geometry gets a quieter first position as a side effect nobody advertises.
The Verdict
Set the nut slots correctly and stop.
That is not a dodge. Compare the two numbers directly: a correct slot depth recovers an error that can run to tens of cents, and a compensated nut recovers something in the low single digits, near or below what most people can hear, at the cost of a nut replacement and — in the Feiten case — a tuning procedure you have to remember for the life of the instrument.
The compensated nut is worth it for a narrow group: players layering open-chord parts against keyboards or against other guitars in a session, where two instruments a few cents apart beat against each other audibly even when neither is wrong on its own. If that is your situation, the per-string mechanical approach is the more honest engineering, because it corrects with geometry rather than by asking you to tune wrong on purpose.
For everyone else the sequence is: new strings, correct slot depth, remeasure, and look at your own six numbers. If they come back at three cents, you have a well-built guitar and an ear that is now calibrated to notice equal temperament, which no nut will fix.
