The Taxi Horn Problem in Gershwin's An American in Paris

The taxi horns in George Gershwin's An American in Paris are one of the most famous novelty effects in the orchestral repertoire. They are also one of the most confusing.

Most percussionists who have played the piece know the basic issue: the score labels the taxi horns as A, B, C, and D, but those letters are not meant to be read as literal pitch names. Modern discussions of the part usually point to a revised pitch collection instead:

• A3, A♭4, B♭4, and D5

That pitch collection is essentially correct. But it leaves out the practical question that matters most in rehearsal:

• Which horn is A? Which horn is B? Which horn is C? Which horn is D?

That was the issue I ran into recently while performing An American in Paris with the Warren Symphony Orchestra. I was playing snare drum, not the taxi horns, but the taxi horn material appeared as cues in my part. The parts we were using were labeled as edited and revised by F. Campbell-Watson. In that performance situation, we knew that the commonly cited "revised pitches" existed, but the relationship between those pitches and the score labels was still unclear.

After comparing the score labels against a Gershwin-supervised recording, the answer became much clearer:

In other words, the revised pitches are not wrong. The missing piece is that they do not go in alphabetical or ascending order.

This article is not a claim about the latest critical edition of An American in Paris. I do not currently have access to the most recent critical edition or rental-only editorial materials, so I cannot verify what that edition says about the taxi horns.

This article is instead a practical source comparison based on the materials available to me:

The point is not to make a final editorial ruling. The point is to document what these available sources suggest and to clarify a problem that performers actually face.

The score uses the labels A, B, C, and D for the taxi horns. At first glance, it is easy to assume those labels are pitch names. They are not.

The revised pitch collection most players encounter is:

But that still leaves an ambiguity. A percussionist does not only need to know which four horns to bring. They need to know how to label them for the part.

The practical question is:

Based on the Gershwin-supervised recording, the answer appears to be:

That means the low A3 horn, the "big horn," is labeled D in the score.

The first important clue comes shortly after rehearsal 3.

At seven measures after rehearsal 3, the score labels the taxi horn as A. In the Gershwin-supervised recording, the pitch sounds as A♭4.

That is significant because it rules out two possible assumptions:

Then at seven measures after rehearsal 4, the score labels the taxi horn as B. In the recording, it sounds as B♭4.

So at this point, the emerging map is:

The passage at rehearsal 5 confirms this relationship more clearly.

At rehearsal 5, the first four measures alternate between horns labeled B and A:

This is not an isolated accident. The A/B relationship is consistent across a repeated passage.

The next important clue appears at rehearsal 5, measure 7. The score labels the horn as C, but the recording sounds like D5.

That gives us the next part of the map:

This is the tiny, high taxi horn.

That matters because it contradicts a simple alphabetical reading of the revised pitch collection. If someone assumed that the four revised pitches should be assigned in order, they might expect C to be one of the middle horns. But in the recording, C is the high D5 horn.

The low horn appears clearly at seven measures after rehearsal 6. The score labels the horn as D, but the recording sounds as A3.

This is the big horn.

That gives the complete working map:

This is the key point. The low A3 horn exists in the revised pitch collection, but it is not "Taxi Horn A" in the score. It is Taxi Horn D.

At rehearsal 7, the score alternates between D and C across the first four measures:

This confirms the C/D relationship:

The score is not saying "C horn, then D horn" in pitch order. It is naming two specific horn identities. In the recording, C is the tiny horn and D is the big horn.

The passage beginning four measures before rehearsal 15 is where the problem becomes more complicated.

In the score, the taxi horn is written on a one-line staff without A/B/C/D labels. The same appears to be true in the original manuscript evidence I have seen: it says taxi horn, but it does not specify which of the four horns should play.

In the Gershwin-supervised recording, what I hear in that location is something like:

But there is a major caveat: those are also the trombone notes.

That means I am not confident that taxi horns are actually sounding there. It is possible that the taxi horn passage was cut. It is also possible that the horns are buried, or that Gershwin preferred the trombone harmony to speak clearly instead of blurring it with taxi horn color.

This passage should not be treated the same way as the clearly labeled passages earlier in the piece.

A modern combined percussion part available on IMSLP assigns this passage as:

That is a practical solution because it keeps the passage within the existing four-horn setup. However, if we apply the label map from the recording, that would produce:

That does not match the low trombone-like contour I hear in the old recording.

In the F. Campbell-Watson edited/revised materials I used with the Warren Symphony Orchestra, the same section appeared in my snare drum part only as cues, and it was notated as an A/B alternation. That is different from the IMSLP combined part, and neither source fully resolves what should happen in performance.

So this passage should be treated as a source conflict:

The responsible conclusion is not "the correct answer is obvious." The responsible conclusion is that the passage before rehearsal 15 requires an informed performance decision.

If I were preparing this part, I would not introduce extra mythical horns like B3, B♭3, and A♭3 based only on this passage. Since those pitches are also present in the trombones, and since the taxi horns are not clearly distinguishable in the recording, I would treat the passage cautiously and coordinate with the conductor, librarian, and available performance materials.

The passage starting four measures before rehearsal 19 brings us back to clearer evidence.

The score alternates between C and D across four measures:

This confirms the same relationship from rehearsal 7:

Again, C is the tiny horn and D is the big horn.

The final major confirmation comes at rehearsal 72.

In the first measure of rehearsal 72, I hear the A♭4 horn. In the second measure, I hear the B♭4 horn. The following material continues to line up with the label map established earlier, including the fact that C is the tiny D5 horn and D is the big A3 horn.

That makes rehearsal 72 important because it confirms the system later in the piece. The labels are not randomly inconsistent. The confusing part is that the labels do not correspond to the pitch names or to a simple ascending order.

The working system is:

If I were labeling the horns for performance based on this source comparison, I would label them this way:

Or, arranged from low to high:

That is the part performers need to know.

The revised pitch collection is correct, but the labels have to be assigned correctly. If the section simply gathers horns pitched A3, A♭4, B♭4, and D5 without knowing which horn corresponds to A, B, C, and D in the part, the players can still end up confused in rehearsal.

The confusion persists because several different problems get collapsed into one.

First, the score labels look like pitch names, but they are not pitch names.

Second, the revised pitch collection is often discussed without clearly explaining the score-label map.

Third, some passages are labeled clearly while others are written on a one-line staff without A/B/C/D labels.

Fourth, practical performance parts do not always agree with each other. The IMSLP combined part and the F. Campbell-Watson edited/revised Warren Symphony materials point to different practical solutions in at least one ambiguous passage.

Finally, the Gershwin-supervised recording is useful, but even that recording does not solve every question. In some passages, especially around rehearsal 15, it is difficult to tell whether the taxi horns are sounding at all or whether the ear is simply catching the trombone line.

The taxi horn problem in An American in Paris is not that the revised pitch collection is wrong. The four practical pitches appear to be:

• A3, A♭4, B♭4, and D5

The real issue is that those pitches do not correspond to the score labels in alphabetical or ascending order.

Based on the Gershwin-supervised recording and the accessible performance materials I have compared, the practical label map is:

That means the "big horn" is D, not A. The "tiny horn" is C.

For performers, that is the useful takeaway. Bring the revised pitch set, but do not assume the pitch order tells you the score labels. Label the horns according to the part:

• A = A♭4

• B = B♭4

• C = D5

• D = A3

The remaining ambiguous passages, especially the unlabeled material before rehearsal 15, should be treated as performance-practice decisions rather than settled pitch facts. The best approach is to coordinate with the conductor, librarian, and edition being used.

The taxi horns are not just a novelty effect. They are a small but revealing example of how messy orchestral performance materials can become when notation, revision history, practical parts, and primary-source recordings do not line up cleanly. For percussionists, the job is not only to play the right sound. Sometimes the job is to figure out what "right" means before the first rehearsal.