The metronome marks for Beethoven’s Ninth Symphony in context

In recent years, Beethoven’s metronome marks for his Ninth Symphony have experienced a renewed relevance, with several ensembles incorporat-ing the indicated speeds in their performances. Nevertheless, previous research has shown that some of the marks have been incorrectly transmitted, and there is the suspicion that further mistakes are still undiscovered. Focusing particularly on the second and fourth movements, this article discusses the historical sources and scholarly contexts for these markings, within a historical framework that draws on Beethoven’s general tempo principles, as well as observations from contemporaries. The article suggests that the trio of the second movement has three speeds that can be justified historically, although the most popular option, minim = 160, arguably has the least supporting evidence. The discussion also draws attention to another metronome mark for the final section of the symphony that has been often overlooked. Finally, it argues that one of the most often cited examples of an erro-neously transmitted metronome mark, the dotted minim = 96 for the Schreckensfanfare found in the later sources, is consistent with Beethoven’s wider practice, and should provide an incentive for performers to experiment with historically informed tempi in this familiar repertory.

C live Brown's 1991 article in this journal described a novel development in the performance of Beethoven's symphonies by early music ensembles, which included following Beethoven's metronome marks. 1 This was nevertheless not altogether new, as the roots of this trend can be traced all the way back at least as far as the performances by Hermann Scherchen and the Orchestra of the Viennese State Opera in the 1950s, 2 or-depending on how strictly the criteria are applied-even to some of the recordings by Felix Weingartner from the 1920s. 3 The early music movement, however, had only just begun to perform Beethoven's symphonies, and Brown identified Roger Norrington and the London Classical Players as the most prominent performers of this kind who engaged with Beethoven's metronome marks. 4 From this, Brown drew the conclusion that many of the indicated speeds which some considered to be impossibly fast were in fact feasible, and that on the whole these marks were representative of what Beethoven had in mind, a few misprints and errors in transmission notwithstanding. Other scholars, including Sandra Rosenblum and Barry Cooper, 5 also argued that Beethoven's metronome marks provide valuable information about performance practice, and they are often printed in scholarly editions as a guide to help clarify the composer's intentions. 6 Much has changed since Brown's article, and performances of the symphonies at or close to the metronome marks have become increasingly common in the early decades of the 21st century. 7 There has also been a significant pushback against some of the attitudes underpinning these performances. Despite Richard Taruskin's initial enthusiasm for Norrington's use of Beethoven's metronome marks, he also argued that such performances represented not a recreation of the past but a newly created performance style. 8 The core of Taruskin's critique, which found much support, centred upon the observation that the surviving historical evidence for musical practices is generally so ambiguous that there is a perpetual need for interpretative guesswork when attempting to establish historical practices. As Taruskin demonstrated, this guesswork is necessarily influenced by a whole range of tacit assumptions regarding what is musically plausible, desirable, or even possible. Furthermore, Taruskin argued that the intentions of composers were only partially knowable: 'composers do not always express them. If they do express them, they may do so disingenuously. Or they may be honestly mistaken. ' 9 Because of such concerns, there still is an ongoing debate about what Beethoven actually had in mind, particularly for movements or sections of movements in which the Italian tempo indication appears to be at odds with the given metronome mark. 10 The Trio of the second movement of the Ninth Symphony is the most notorious but far from the only example, and strong feelings exist on all sides of this debate, often still fuelled by the kind of rhetoric at which Taruskin took aim. 11 The metronome marks for the Ninth Symphony are neither the most controversial-that honour is reserved for the minim = 138 of the first movement of the Piano Sonata op.106-nor the least controversial in Beethoven's music, as some individual marks have been identified as probably having been incorrectly transmitted. 12 This article will explore the historical evidence for the different This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. pAGE 2 OF 18 Early Music interpretations of the metronome marks in the Ninth Symphony that can be traced back to the composer, by considering the extent to which they correspond to Beethoven's wider tempo principles, as well as contemporary descriptions of early performances.

Sources for the metronome marks of Beethoven's Ninth Symphony
Although there are three metronome marks for the symphony that can be dated to before the first rehearsals and performances in May 1824 in Vienna in Beethoven's presence, the majority of the metronome marks were produced in September 1826. 13 The documented metronome marks for the Ninth Symphony are displayed in Appendix 1; the sources for these marks are as follows, in order of chronology: A: The autograph score of 1823-4, containing the earliest known metronome marks for the first tempo and the penultimate Prestissimo of the symphony (illus.1). B: The engraver's copy of 1824-6, which may have been used in the first performance, and which contains a single metronome mark on the title-page of the second movement (illus.2 appeared in the earlier sources, possibly due to a typesetting error in Source E. 15 Furthermorebased at least in part on a misprinted representation of Source F in Moscheles's biography in which none of the note values has dots, and in which the tempo indications appear without corrections 16 -Baensch argued that Source F was simply copied from Source E and was therefore of no value. The original Source F (illus.6), however, problematizes this interpretation: several tempo indications have been corrected to resemble those in the score, and it is the only source to include the correct note value for the Molto vivace in the second movement, an observation also made recently by Byung-Jun Park. 17 Clearly, Source F is not just a copy of Source E, but the extent to which Beethoven was involved in these corrections, or whether Schindler made them on his own account, remains unknown. Nevertheless, most critical editions have accepted Baensch's reading, and include dotted minim = 66 instead of 96 for the opening Presto of the finale, 18 an issue which will be discussed in detail towards the end of this article.
Two other metronome marks have also been identified as possibly having been incorrectly transmitted. The first of these is the speed for the Turkish march marked Allegro assai vivace alla marcia, which occurs in Sources D, E and F as dotted crotchet = 84. As illustration 3 shows, Karl did not write down the note value in the conversation books, and he probably added it later in the letter to Schott. As Hermann Beck was the first to show in the 1950s, and as many have since agreed, dotted minim = 84 is more in line with other metronome marks by Beethoven for similar movements, and is also more consistent within the context of the finale of the Ninth Symphony, as otherwise two variations of the 'Freude' theme would be at half the speed of the others. 19 The second metronome mark whose accuracy has been questioned is the minim = 116 for the Presto of the trio in the second movement, which Peter Stadlen has discussed extensively. 20 According to Stadlen, this slow speed was in blatant contradiction of the tempo indication, which he took as evidence for a reading of semibreve = 116. Early Music PAGE 5 OF 18 Erica Buurman, however, has argued on the basis of the sketches that minim = 116 is representative of Beethoven's intended tempo. 21 Despite these studies on individual metronome marks and the large number of performers interested in this material, no single contribution has discussed all the metronome marks for this important work in the context of Beethoven's overall practice. The following section discusses Beethoven's writings on tempo from his earliest compositions onwards, and the extent to which these principles can be observed in the metronome marks that he produced prior to 1826.

Principles of Beethoven's tempo and metronome marks
Beethoven's earliest written comments on tempo can found in the sketches of the song 'Klage' , WoO 113, which remained unpublished throughout his life, and of which he drafted two different versions in 1790. On a half-finished draft of the first version, the following statement appears: [1] What follows will be sung even slower, adagio or andante quasi adagio at the most.
[2] Andante in a 2/4 metre has to be taken much faster than the tempo here in this song. [3] It seems that it is impossible that the [second section] remains in 2/4, because it is much too slow.
[4] [It appears] best to set [this section] in .
[5] The first [part] in E has to remain in 2/4, because otherwise it will be sung too slowly.
[6] One would sooner take a slow tempo in the case of long notes than with short ones. [7] For example, with crotchets slower than with quavers. [8] The shorter notes also determine the tempo, for instance semiquavers/demisemiquavers in 2/4 time make it very slow. [9] Perhaps the converse is also true. 22 The first and second sentences indicate that Beethoven intended the minor section to be slower than the tempo at the beginning of the song, and much slower than an Andante in 2/4 would normally be. As the third, fourth and fifth sentences show, his solution to this problem was to write both sections in different metres, which resulted in the major section remaining in 2/4 and the minor section being changed into . Both metres have two beats in every bar, but they are indicated by different note values: in 2/4 by crotchets, and in  by minims. It seems likely that the sixth and seventh sentences refer to the principle that metres with larger note values indicating the beat suggest a slower tempo than those with smaller note values. The minim beat in  can therefore be expected to be slower than the crotchet beat in 2/4, and if the same number of notes per beat is maintained, the music will sound slower. It is for this reason that the minor section, which in the first version contains two quavers in the voice part for every crotchet beat in 2/4, is written with two crotchets for every minim beat in  in the second version. The last two sentences seem to allude not to the note values in the time signature, but to the range of note values that are used in a bar: an increase of smaller note values also implies a decrease in speed. The somewhat ambiguous last sentence perhaps is also best understood in this light: it implies that the chosen tempo also determines the range of note values that the composer can use.
The relevant parts of the first and second version of WoO 113 can be seen in example 1. In the first four bars of the second version, very little has changed beyond some minor alterations in the piano part, but after the double bar at the end of bar 15, Beethoven changed the time signature from 2/4 to , doubled all the note values, and added a tempo indication clearly communicating a slower tempo. These three factors combined-note values, tempo indications and metre-were used by Beethoven to communicate the tempo that he had in mind at this point. This is further supported by musical treatises circulating at the time, particularly Kirnberger's Die Kunst des reinen Satzes (1776), some parts of which Beethoven copied by hand, 23 and which describes tempo in almost identical terms. 24 Furthermore, Beethoven seems never to have abandoned these principles, as they can also be detected in the metronome marks he produced from 1815 onwards for his string quartets, symphonies and other works. Seemingly irrespective of other factors such as instrumentation and genre, the same combination of note values, metre and tempo indication almost always correlates with a similar speed, as Rudolf Kolisch first observed in 1943. 25 To name one example out of many, the fourth movement of the Fifth Symphony op.67 1 Beethoven, Song 'Klage' WoO 113: (a) first version, bars 11-17; (b) second version, bars [11][12][13][14][15][16][17] and the first movement of the String Quartet op.74, both Allegros in  with crotchets, quavers and semiquavers in scale figures, are both marked minim = 84, despite the clear differences in instrumentation and character. Furthermore, Allegros in  with more extensive semiquaver figurations have slower metronome speeds, such as the Allegro risoluto in the finale of the Piano Sonata op.106 marked minim = 72 (written as crotchet = 144). The lack of influence of factors such as the venue or the number of instrumentalists per part on the metronome marks might suggest that they were merely intended as theoretical goals. However, the welldocumented interest that Beethoven expressed in all matters related to tempo, 26 as well as the fact that with very few exceptions all of his metronome marks were set after the work had been performed or rehearsed several times, 27 suggest that his metronome marks must also have had some degree of practical significance.
Although some of Kolisch's subsequent conclusions based on these observations have come under criticism, 28 the broader points were echoed by Rosenblum, Brown and others, who agreed that these principles could be used to make predictive metronome marks for works for which Beethoven did not leave any. 29 Similarly, they can also be used to check whether a metronome mark is likely to be misprinted or otherwise erroneously transmitted.
Nevertheless, this appeal to consistency has its problems: Beethoven was clearly not completely consistent, as passages with the same tempo indication in 2/4 and  typically have the same metronomic speed as long as the number of notes per bar remains the same, in contrast to what his words on the sketch of WoO 113 suggest. 30 This implies that Beethoven's sense of tempo either changed before he started using the metronome in the last twelve years of his life, or that he was not always as consciously aware of his own sense of tempo as his deliberations suggest.
The latter seems the most likely explanation, considering how many examples there are of Beethoven having difficulties indicating his intended tempo. According to Gerhard Wegeler, the finale of the Piano Trio op.1 no.2 was initially written in the wrong metre and with a range of note values that did not reflect the speed that the composer had in mind, something which Beethoven corrected after this was pointed out. 31 (Wegeler's memory probably fails him here in the details: according to him the original metre was 4/4, for which Beethoven always used . A much more likely candidate is , as much of the material in the final version is found in the sketches in note values twice as large. 32 ) Another example of an incorrect metre is found in the well-known letter that Beethoven wrote in 1812 to the publisher of his Mass in C op.86: You will have received the corrections for the Mass ... at the beginning of the Gloria I have written  instead of  and changed the tempo from the original indications [from Allegro con brio to Allegro]. I was seduced into doing this because of a bad performance, during which the tempo was taken too fast. Now, not having seen the Mass for a long time, this caught my attention immediately and I saw that, unfortunately, such a thing must be left to chance. 33 The autograph score confirms that the Gloria was originally marked Allegro con brio and , 34 which Beethoven changed for the first edition after an early performance. When proofreading the first edition, however, he reconsidered this change and evidently wanted to restore the original metre and tempo indication. For an unknown reason, however, this did not happen: in Beethoven's copy of the first edition, which contains some corrections, the Gloria is still marked Allegro and . 35 It is therefore possible that, on some occasions, inconsistent metres and tempo indications ended up in source material. Despite Beethoven's general consistency in his tempo indications, he was clearly not immune to error, and music in duple and quadruple metres seems to have been at particular risk of inconsistent time signatures. Furthermore, as the above examples show, Beethoven occasionally seems to have used note values and tempo indications inconsistently, although the former seem to have been rare and the latter of relatively minor consequence. The following discussion explores whether-on balance of probability-these or other errors may have occurred in the metronome marks in each movement of the Ninth Symphony.
The headings for each of the ensuing sections give the metronome marks that on balance are best supported by the historical evidence.

Movement 1
Allegro ma non troppo e un poco maestoso, 2/4, crotchet = 88 Considering the similarity in the pencil markings between Beethoven's proto-metronome mark '108 or 120 Maelzel' and the barely legible original tempo indication, it seems likely that both were written around the same time. This would put Beethoven's earliest attempt at a metronome mark somewhere near the beginning of 1824, when he was finishing the symphony. At this point, Beethoven was clearly unsure of what the metronome mark and perhaps even the tempo indication should be, but the two numbers he was thinking about at this point are close to the speeds given in 1818 to two of his other Allegros without ma non troppo in the same metre that share some characteristics with this movement. The first is the last movement of the String Quartet op.18 no.1, marked Allegro and crotchet = 120, which contains triplet semiquavers throughout but no extensive demisemiquaver figuration; both the tempo indication and the range of note values employed suggest a faster tempo than in the Ninth Symphony. The second is found in the first movement of the String Quartet op.18 no.2, marked Allegro and crotchet = 96 with a range of note values comparable to those in the first movement of the Symphony. Thus there was at least a precedent for Beethoven's earliest speeds for this movement, especially if one considers that the ma non troppo could very well have been a late addition to the tempo indication. Either way, considering the similarities with these string quartet movements, there is no reason to believe that the speed of crotchet = 88 for the first movement of the symphony is based on an error, nor is there any evidence for the recent assertion by Almudena Martin-Castro and Iñaki Ucar that the proto-metronome marking is a result of Beethoven's confusion with reading his metronome. 36 Most recent recordings, however, even those that print the metronome marks in the booklet such as Norrington's recording with the London Classical Players, take a somewhat slower speed, with the notable exception of Benjamin Zander's 2017 interpretation with the Philharmonia. 37

Movement 2: Scherzo
Molto vivace, 3/4, dotted minim = 116. Trio: Presto, , minim or semibreve = 116 The Scherzo consists of crotchets all the way through, with single quavers forming a recognizable rhythmic motif. The metronome mark of dotted minim = 108 on the engraver's copy of 1824 seems close enough to eliminate the possibility of a misprint, although in 1826 after the experience of the first performances Beethoven evidently thought this movement would work better ever so slightly faster. Our confidence in this speed can be further bolstered by comparing it to the marks for similar movements. This is the only time, unfortunately, that Beethoven used Molto vivace in 3/4 without adding other tempo indications, so the best option is to compare this section to Allegros and Prestos in 3/4 with the same range of note values, with the expectation that the speed for this movement falls somewhere in the middle. This is indeed the case: the third movement of the Sixth Symphony (an Allegro, with some small groups of quavers) is marked dotted minim = 108, as are several other Allegros with similar characteristics; 38 and the third movement of the Seventh Symphony (a Presto, with only crotchets, and consequently a bit faster) is marked dotted minim = 132. All in all, the speed of dotted minim = 116 for a tempo marking of Molto vivace with a few single quavers seems consistent, and it is often adhered to or approximated in recordings and performances, even by conductors who ignore Beethoven's metronome marks elsewhere. 39 The intended speed for the Trio has caused considerable controversy, and generally three different candidates have been advanced: semibreve = 116, minim = 116, and minim = 160. All three of these can be substantiated by reasonable arguments, which nonetheless do not fit all the available evidence precisely. It is worth briefly examining the arguments for and against each of these, starting in reverse order.
Around the turn of the 20th century, Weingartner suggested a speed of minim = 160 for the Trio as a compromise between the suspiciously fast semibreve = 116 and the rather slow minim = 116. 40 About a century later again, Cooper and Jonathan Del Mar offered a possible historical justification for the mark of 116 recorded in Source C, by proposing that Beethoven was operating the metronome and dictating to Karl, and that the latter misheard Beethoven, mistaking einhundertsechzig (160) for einhundertsechzehn (116). 41 But that assumes that this mistake went completely unnoticed by both Beethoven and Karl. This seems unlikely, as the latter was able to spot relatively small discrepancies between speeds with comparative ease, evident in Karl's entry in the conversation books referring to the final Prestissimo of the fourth movement: 'you take it faster than 120. 132. That is how we had it in the morning.' 42 It therefore seems unlikely that Karl would have missed a mistake of this magnitude in a passage in which the rhythmic pattern is fairly straightforward. Furthermore, since Karl was the one telling Beethoven which speed he was taking, Beethoven was behind the piano and Karl was establishing the speeds with the metronome when this entry was made. So, unless the setup was different when Karl wrote down minim = 116, Beethoven would not be calling out numbers to his nephew, which in turn makes it unlikely that the metronome mark is based on Karl's mishearing.
There may nevertheless be a certain historical validity to taking the Trio at minim = 160. A review of the first Viennese performance of the symphony states that Beethoven indicated the tempos for each section, with Michael Umlauf, the conductor who had previously successfully premiered several of Beethoven's works, presumably following him. The review describes the Trio as 'a brilliant march' , 43 a description which Brown has argued is more easily associated with semibreve = 116 than with minim = 116. 44 The metronome marks for Beethoven's marches, however, generally indicate a beat of around 80, although the note value can vary, as the three metronome marks by Beethoven for his marches indicate: quaver = 80 for the funeral march in the Third Symphony (Adagio assai), quaver = 76 for the march in the Septet op.20, and dotted minim = 84 for the Turkish March in the Ninth Symphony. Furthermore, several of the marches in compositions that Czerny studied with Beethoven received editorial metronome marks around 80, 45 including the march in the Choral Fantasia (Vivace, crotchet = 80) and the second movement of the Piano Sonata op.101 (Vivace alla Marcia, minim between 76 and 84). 46 Assuming that the reviewer had heard marches by Beethoven, the description 'a brilliant march' could be taken as evidence that the Trio proceeded at semibreve = 80 or minim = 160 during the first performance.
The evidence suggests, however, that Beethoven changed his mind in 1826, with the number 116 applying to either a minim or semibreve. The arguments in favour of minim = 116 include the fact that Sources D, E and F contain this speed, and the observation that the alternative of semibreve = 116 is so fast that it seems implausible. 47 Evidence in the source material may also support this reading. As Buurman has shown, Beethoven initially notated the section that would become the Presto not in  but in 2/4, and that this was changed only relatively late in the creative process, during which Beethoven experimented with a precise tempo relationship between the 2/4 of the Presto and the 3/4 of the Molto vivace. This suggests that at least in its early conception three crotchets in the latter would fit in the same amount of time as two in the former. 48 Furthermore, in 2/4, minim = 116 seems more in line with other markings in that metre, especially considering the fact that the sketches also imply a connection with the Allegro molto second movement of the Piano Sonata op.110, which at the start has a similar range of note values and for which Czerny and Moscheles suggested speeds close to minim = 116. 49 The connection with the second movement of op.110, however, is ambiguous, both because it was established relatively early in the creative process and because unlike the Trio the sonata includes extensive quaver figuration. Other movements with a similar speed and tempo indication, such as the Presto finale of the Septet op.20 (minim = 112), also tend to have more extensive quaver figuration. 50 But it may be that at that stage in the creative process, Beethoven was considering using more quaver figurations than ended up in the final version, and based on these early designs, minim = 116 seems a reasonable interpretation.
An early review may also support this reading. The second performance of the symphony on 23 May 1824 was described by Friedrich August Kanne in the following way: 'The grotesque leaps that Beethoven's genius makes in the Scherzo at hand are often of such a bold nature, and are executed with such rapid power, that one readily understands how he could mix an Allabreve into this tempo, in which the ear at once regains new strength' . 51 Kanne's comments may suggest a somewhat more relaxed pace, although their meaning is ambiguous, and, as David Levy has argued, neither Kanne's review of the second performance nor the above-cited review of the Viennese premiere can be taken as definitive descriptions of the tempo taken at these performances. 52 The alternative interpretation, of semibreve = 116, was first proposed by Stadlen and echoed amongst others by Stewart Young. 53 Notwithstanding the aforementioned objections, its practicality in performance is indicated by Zander's recent recording at that speed. 54 Arguments for this tempo interpretation include the stringendo leading up to the Trio, which would suggest a faster speed; the possibility that the note value was incorrectly entered from the start; and the fact that the word Prestissimo appears in pencil on the autograph of the trio, suggesting an utmost degree of speed. There are indeed several metronome marks by Beethoven for other prestos in this metre that are close to or seem consistent with a faster speed for the Trio. The final movement of the String Quartet op.59 no.2, for instance, contains more quaver figuration than the Trio, and accordingly has a speed of semibreve = 88; the Più Presto coda, which contains fewer quavers and is more similar to the Trio, is marked semibreve = 112. Similarly, the last movement of the Fifth Symphony ends with a Presto with only crotchets, marked semibreve = 112. Overall, semibreve = 116 does seem more consistent with Beethoven's wider output, despite its contradiction with the evidence from the sketches.
In conclusion, the earlier and more private evidence of the compositional process seems to suggest a reading of minim = 116 on the basis of tempo relations in the sketches, while the later and more public evidence of the description of the first performance and the consistency with Beethoven's wider output suggests readings of minim = 160 and semibreve = 116, respectively. Attempting to winnow down those options to a single one would suggest a greater certainty than the surviving evidence will reasonably support, so it seems most prudent to let performers decide that by which they want to be informed.

Movement 3
Adagio molto e cantabile, , crotchet = 60. Andante moderato, 3/4, crotchet = 63 This movement, too, has a somewhat peculiar tempo relationship between two sections in a different metre, as the crotchet pulse hardly changes between the Adagio molto and Andante. On the basis of this, one might suspect an error, but the crotchet = 60 for the Adagio molto is consistent with the two other Adagios in this metre that Beethoven gave metronome marks in 1817 and 1818. First, the opening Adagio molto of the First Symphony is marked quaver = 88, but it has much more extensive quaver and semiquaver figuration from bar 5 onwards, so a slower speed is expected. Second, the Molto adagio of the second movement of the String Quartet op.59 no.2 is also marked crotchet = 60, and includes a very similar range and distribution of note values (minims, crotchets and quavers at the start, and smaller note values later).
Beethoven's Andantes are somewhat less suitable to compare, as the contemporary definitions of the term suggest that it encompasses a variety of different kinds of tempo. 55 Nevertheless, the speed for the Andante moderato seems consistent with the only other Andante in 3/4 with a metronome mark by Beethoven, which is the Andante con moto marked crotchet = 69 at the beginning of the String Quartet op.59 no.3 with semiquavers in every instrumental part. The editorial speeds by Czerny and Moscheles for the only Andante in 3/4 in the piano sonatas are also in a similar range: the Andante molto cantabile ed espressivo in the Piano Sonata op.109 is marked between crotchet = 63 and 72, 56 although this section generally has larger note values at the beginning. There are no Adagios in 3/4 with the same range of note values with a metronome mark by Beethoven, but all of those for which Czerny and Moscheles gave editorial speeds are slower than crotchet = 50. 57 So Beethoven's speed for the Andante moderato seems consistent.
The section immediately after the march is marked Allegro ma non tanto minim = 120 in , and contains mostly quavers at the start and more crotchets later. There are no other sections in Beethoven's oeuvre with the same tempo indication and a metronome mark, yet there are somewhat faster metronome marks for some of the Allegros with the same range of note values but without further qualification. These are the minim = 138 for first movement of the Piano Sonata op.106 and the minim = 132 (printed as semibreve = 66) for the fourth movement of the String Quartet op.18 no.4. Again, there is no reason to assume an error here.
The Prestissimo that follows on the list-the short Poco adagio evidently was not given a metronome mark, nor was the Stringendo passage that connects the two-was marked Presto in the autograph score, with the word Prestissimo crossed out, as illustration 1(b) shows. These changes, which Del Mar speculates took place during the rehearsal, seem to suggest a possible change of mind on Beethoven's part. As the discussion of the Trio in the second movement showed, compared to other Prestos in , the combination of mostly crotchet and quaver figuration and a speed of minim = 132 is an outlier. But as Brown suggested, 58 it may be that the use of  here is inconsistent, and  seems much more consistent with other sections with these characteristics, particularly in light of the first statement of the 'Freude' theme marked Allegro assai, minim = 80 in  with a similar range of note values. 59 Performers who follow the metronome marks usually play the final Prestissimo at the same speed as the abovementioned Prestissimo marked Presto in the autograph. There is some evidence, however, that Beethoven may have intended a faster speed when he came to set the metronome marks: as illustration 3 shows, at the end of the list of metronome marks Karl wrote '88' , which Del Mar has hypothesized may have been the intended speed for the final section of the symphony, overlooked in all subsequent sources due to the confusing way it was written down. 60  The Schreckensfanfare All metronome marks of Beethoven's Ninth Symphony discussed so far can be seen as consistent with or at least not contradicting those in the rest of Beethoven's oeuvre, with the exception of the misprinted speed for the Turkish March, and the Trio of the second movement, for which the evidence is inconclusive. With this in mind, it is time to revisit the opening Presto of the Schreckensfanfare. As discussed at the beginning of this article, in the conversation book, presentation copy, and letter to Schott-all in Karl's hand-the speed for this section is dotted minim = 66, while all subsequent sources have dotted minim = 96. Baensch argued that the cause of this was a misprint by Schott which was copied into the other sources, and that the original dotted minim = 66 corresponds to the speed that Beethoven had in mind, an interpretation that is echoed in almost every discussion since. 64 However, despite this scholarly consensus, Baensch's interpretation is less plausible if one considers the principles of Beethoven's tempo indications. As stated above, if two pieces by Beethoven have the same tempo indications, metre and range of note values, they generally have the same speed, regardless of other factors. If the metre and tempo indication are the same but one has smaller note values than the other, the former will have a slower speed, and vice versa. With this in mind, a number of notable inconsistencies can be observed.
First, compared to all of Beethoven's metronome marks for Prestos in the same metre, dotted minim = 66 is suspiciously slow: the third movement of the String Quartet op.74 in quavers throughout is marked dotted minim = 100; the subsequent Presto quasi prestissimo in crotchets throughout has a speed of dotted semibreve = 100, which is three times as fast as dotted minim = 66 with note values only twice as large; the Presto of the Seventh Symphony-mainly in crotchets, so expected to be somewhat faster-is with dotted minim = 132 exactly twice as fast as the speed in the conversation book. The subsequent Assai meno Presto contains fewer quavers than the Schreckensfanfare, but with dotted minim = 84 is still significantly faster despite its clearly slower tempo indication. So dotted minim = 66 for a Presto in 3/4 with quavers is unmistakeably something of an aberration compared to other Prestos in Beethoven's oeuvre.
Second, other sections with a speed of about dotted minim = 66 in 3/4 with quavers as the most common note value, such as the third movement of the String Quartet op.59 no.2 with a speed of dotted minim = 69, are generally marked Allegretto rather than Presto. Third, even Beethoven's metronome marks for sections marked Allegro with quavers are a lot faster than dotted minim = 66: the Più Allegro that follows the Allegro assai vivace ma serioso in the third movement of the String Quartet op.95 is marked dotted minim = 80; the third movements of the String Quartet op.18 no.3 and the Second Symphony are both marked dotted minim = 100; and the third movements of the Sixth Symphony and the String Quartet op.18 no.1 are marked dotted minim = 108. Although these Allegros do not have constant quaver motion in the way that the Schreckensfanfare has, the figuration clearly has an effect on the overall tempo, as Allegros without quavers at all are faster still, such as the scherzos from the Third Symphony (Allegro vivace, dotted minim = 116) and the Septet op.20 (Allegro molto e vivace, dotted minim = 126).
All of this makes dotted minim = 66 for a Presto with quavers rather suspect, and the dotted minim = 96 that ended up in the published sources more likely to be what Beethoven had in mind. But this raises the question how dotted minim = 66 ended up in the conversation book in the first place. A possible explanation could perhaps be found in the setup when the marks were produced. As previously stated, it is unlikely that Karl had the score of the symphony in front of him, as otherwise he would have written down the speeds there straight away and not in the conversation books. (As Stadlen argued, the presentation copy was probably out for binding at this time. 65 ) So it seems probable that Beethoven, possibly seated at a piano, had the autograph score, and Karl had the metronome and the conversation book. In such a set-up, it would have been particularly difficult to provide a metronome mark for this Presto in dotted minims: the syncopated rhythms and the offbeat leaps Early Music PAGE 13 OF 18 in the melody distort the sense of a 3/4 metre filled with quavers, as shown by (a) in example 2. In fact, Karl could have easily heard a 3/4 bar filled with triplets, as indicated by (b) in example 2, which would have resulted in a metronome mark for a Presto that is a third slower than what Beethoven played. It therefore could be hypothesized that the initial dotted minim = 66 was based on an error by Karl, copied by Karl into the presentation copy and the letter to Schott along with the other errors, and that Beethoven sent Schott the number 96 as a correction later. Although such correspondence has never been found, it would be easy for a short note with a single correction to disappear. The alternative explanation, that Beethoven was uniquely inconsistent in this Presto, only for Schott to misprint it at a more consistent speed, relies on an extraordinary coincidence, and therefore falls foul of Occam's Razor.
Thus, on balance of probability, dotted minim = 96 seems much more likely than dotted minim = 66. Given the scrutiny to which other metronome marks in the Ninth Symphony have been subjected, the fact that Baensch's explanation has gone unchallenged is remarkable. Principally this may reflect the prevalent view of performers and scholars that dotted minim = 96 is unfeasibly fast. The earliest recordings of Beethoven's Ninth Symphony, which had just started to become available when Baensch was doing his work, generally stay far below dotted minim = 96. 66 More recently, Levy has claimed that 'there can be no doubt that 96 is a mistake, as the music is absolutely unplayable at this speed' , 67 and Zander has described the speed as 'obviously preposterous [and] truly ridiculous' . 68 On the other hand, no 19th-century commentators considered dotted minim = 96 to be too fast for the Schreckensfanfare. George Grove, despite condemning semibreve = 116 in the Presto of the second movement as 'almost impossible for the horns' , stated that 'Beethoven's care that all the indications of tempo &c should be fully given in his published works was as minute and unfailing as usual' , and left the dotted minim = 96 for the Schreckensfanfare otherwise unremarked upon. 69 Weingartner went so far as to sanction the speed explicitly, writing that 'the metronome mark dotted minim = 96 is too fast for the bass-recitative, although not for the "Fanfares"' , for which he furthermore recommended 'the quickest [speed] which is compatible with a continuous fortissimo' . 70 So although that speed may seem overly fast, and more aspirational than an actual goal to be achieved, there are historical reasons for at least attempting it, and there is no historical reason to limit the speed to dotted minim = 66, as many have done. However, a conductor may prefer the slower speed for his or her own artistic reasons, or, as Weingartner also implied, because it is the maximum attainable with the forces at hand.
Finally, there is the issue of the recitative that follows the fanfare, marked Selon le caractere d'un Recitative mais in Tempo. Entries in Beethoven's conversation books during the time of the rehearsals for the first Viennese performance in 1824 indicate that the recitative was also 'tremendously difficult' to play in the tempo that Beethoven had in mind, 71 which seems to have been somewhat slower than the metronome mark for the previous section. This was first stated by Leopold Sonnleithner, a friend of Beethoven who attended many of the rehearsals for the first Viennese performance, and who specified that Beethoven wanted the recitative played at 'a rapid pace, that is not presto, but also not andante ' . 72 If that is accurate, one would assume a speed somewhere near that for an Allegretto in 3/4 with crotchets and quavers, which as shown earlier is estimated to be around dotted minim = 66, although Weingartner's somewhat slower suggestion of crotchet = 168 (dotted minim = 56) would also fit this broad description. Although there appear to have been no recorded attempts to approach dotted minim = 96 in the Schreckensfanfare, there are several that play both the fanfare and the recitative at around dotted minim = 66, including René Leibowitz, Norrington and others. 73

Conclusions
The discussion of existing scholarship and the comparisons between Beethoven's metronome marks for his Ninth Symphony and those for similar movements allow several observations to be made. The speeds for many sections of the symphony are remarkably close to the speeds Beethoven gave similar movements approximately ten years earlier, suggesting that-at least subconsciously, if not in practice-he maintained certain underlying principles when it came to musical speed. As with several earlier works, there is some evidence that Beethoven may have used time signatures inconsistently, and on the basis of the abovementioned comparisons, it seems likely that this happened in the case of the first Prestissimo in the finale in , which might make more sense in . Also, as previous scholars already established, these comparisons show that the note value for the metronome mark of the Turkish March was incorrectly transmitted. Neither Norrington, nor Baensch, nor the editor of the recent Henle edition spotted the erroneous note value for the Turkish march, despite their familiarity with the sources. This implies that Beethoven's intended tempo can be counterintuitive, something attested to by the controversies surrounding the topic. Furthermore, the metronome mark at the opening of the Schreckensfanfare is inconsistent with other Prestos in the sources in Karl's hand, but consistent in all later sources, so it seems that an error is likely there too. Since the sources in Karl's hand were not studied until the early 20th century, for the first century of the symphony's performance history, dotted minim = 96 was the only speed in any of the editions, and was not considered problematic by musical commentators. Finally, there may also be a hitherto unnoticed metronome mark for the final section of the symphony.
None of the above is intended to be followed rigidly or dogmatically; 74 instead, it is an attempt to explore the options that can be most plausibly justified on the basis of the historical evidence. Radically changing the speed of any music poses technical challenges, and might not immediately lead to positive results, something to which many performers can attest. Nevertheless, a greater understanding of how Beethoven may have thought about tempo can inspire the making of modern historically informed artistic decisions that would otherwise probably not even be considered.

The metronome marks for Beethoven's Ninth Symphony in context
In recent years, Beethoven's metronome marks for his Ninth Symphony have experienced a renewed relevance, with several ensembles incorporating the indicated speeds in their performances. Nevertheless, previous research has shown that some of the marks have been incorrectly transmitted, and there is the suspicion that further mistakes are still undiscovered. Focusing particularly on the second and fourth movements, this article discusses the historical sources and scholarly contexts for these markings, within a historical framework that draws on Beethoven's general tempo principles, as well as observations from contemporaries. The article suggests that the trio of the second movement has three speeds that can be justified historically, although the most popular option, minim = 160, arguably has the least supporting evidence. The discussion also draws attention to another metronome mark for the final section of the symphony that has been often overlooked. Finally, it argues that one of the most often cited examples of an erroneously transmitted metronome mark, the dotted minim = 96 for the Schreckensfanfare found in the later sources, is consistent with Beethoven's wider practice, and should provide an incentive for performers to experiment with historically informed tempi in this familiar repertory.