Metallica’s song “Master of Puppets” has extremely powerful and driving momentum, which is especially impressive given that it is riddled with meter changes and thrashes on for almost nine minutes. The 1985 album of the same name, on which the song was released, was praised by all sorts of critics, and is commonly described as one of the best metal albums ever released. ((See Steve Huey’s review of this album for All Music Guide, http://www.allmusic.com/album/master-of-puppets-mw0000667490 )) The title song in particular is a fan favorite, and is a staple of the band’s live sets. “Master of Puppets” is also one of Metallica’s most popular songs among aspiring guitarists, and the number of home-made transcriptions of this song available online is simply staggering — for example, ultimate-guitar.com has at least 45 separate tabs of this song alone, not including tabs of the whole album. ((At the time I’m writing this, 911tabs.com lists literally hundreds of “Master of Puppets” tabs hosted on other webpages, but a lot of those are redundant copies of the same transcriptions.))
But I feel kind of funny every time I look at a transcription of “Master of Puppets.” There’s a lot of changing meter in this song, and there’s one riff in particular in the verses which has several measures of 4/4 followed by what is usually transcribed as a single measure of 5/8.
from a transcription published in 1988 by Cherry Lane Music Company, copyright owned by Creeping Death Music
The problem is that on the album version of the track and in some live recordings, it doesn’t really feel like an even 5/8 — the rhythm is consistently off enough that if you try to tap out really strict eighth notes, you end up not being with the band when you get through to other side. There’s a lot of research literature within the fields of music theory and music cognition which focuses on such “microtiming deviations,” but most of this research deals with rubato (slowing down and speeding up in the course of a phrase) or differences in timing between a solo instrument and the accompanying ensemble. ((In some styles of jazz, especially, the lead player can be as much as several beats off from the rhythm section. See Richard Ashley’s article “Do[n’t] Change A Hair For Me: The Art of Jazz Rubato” in Music Perception Vol. 19, No. 3 (Spring 2002). )) That doesn’t seem to be what is going on here.
To get a better understanding of what was happening, I opened up the album version of the song in Audacity (a free sound-editing program) and started measuring the timings of Metallica’s performance. In the spectrum viewing mode, it is very easy to see the drum articulations in most rock recordings, and with close zooming it is fairly easy to place a label on each important beat with an accuracy around 2 or 3 thousandths of a second. The status box at the bottom of the window tells you the exact timing of this label down to more decimal places than could possibly be practically useful. Subtracting the timing of one label from the next previous label calculates what music scholars call the “inter-onset interval,” in other words, the duration between the start of one note and the start of the next. This method makes it easy (although a bit time-consuming) to closely study the rhythmic feel of any recorded performance with drums or some other instrument with clear articulations, such as piano or guitar. ((This method is notoriously unhelpful for measuring sounds made by the human voice or bowed string instruments, which have less sharp attacks. ))
I wanted to be thorough, and measure enough timings of this pattern to get a good average, but I do still have a life outside of music research, so I only looked at the occurrences of this pattern in the first verse. That still gave me eight instances of the 5/8 pattern to study. I measured the surrounding quarter note beats for comparison, which are mostly clearly marked by the drum pattern. ((In all but the first two cases, the boundary between the first and second duration in my chart was not clearly visible. For the latter six rows of the chart, this combined half-note duration was exactly twice as long as the quarter-note duration I expected, so I gave the first two columns half the value of the larger measurement. Although some people might argue I’ve “fudged the numbers,” it doesn’t affect the result that the second duration measurement of the 5/8 measure consistently is longer than the average quarter note. )) If you look at the chart below, there is a clear and surprisingly consistent pattern to how Metallica performs this rhythm.
Timings from the first verse of the album version of “Master of Puppets.” Each column corresponds to a note or pair of notes with a red number beneath it in the above transcription.
As you can see, Metallica’s timing keeps pretty consistently to .15 seconds for an eighth note and .29 seconds for a quarter (or two eighth notes), except for the middle of the 5/8 measure. After the first three eighth notes of this measure, you can hear a brief pause before the last two eighth notes, a pause which is almost always .04 or .05 seconds (about a third of an eighth note), and which makes the measurement of these two eighth notes grouped together .34 or .35 seconds. What makes this rhythmic idiosyncracy different from what has been studied by most music theorists is that this slightly attenuated beat is performed by the whole ensemble in unison, and it’s not a delay that is “made up for” right afterwards. In other words, it’s not a local deviation from the beat that maintains the pulse over a longer span of music, but a permanent shift of where the beat occurs.
This 5/8 measure deliberately disrupts the song’s pulse as much as possible. Even if the band played the eighth notes in straight timing, the quarter note pulse and half note pulse would both be disrupted by the odd length of the 5/8 measure. The 5/8 measure places accents on the second and fourth eighth notes, against the pulse of the preceding 4/4 measures, but then the following measures continue to reinforce this new location of the beat. ((A couple of scholars have observed this “new pulse location” effect before me, especially Glenn T. Pillsbury in his 2006 book Damage Incorporated: Metallica and the Production of Musical Identity. ))
For a long time, I thought that the transcriptions I’d seen were all wrong, that this riff did not have a 5/8 measure in it, but until I measured it I wasn’t sure if my ears were fooling me or not. I still think this is a valid question; 5/8 is “supposed to be” five equally timed eighth notes, so does this performance of “Master of Puppets” count as 5/8 when one eighth note is regularly 30% longer than the rest?
I feel like that isn’t a question that can really be answered completely — on the one hand, 5/8 is probably the easiest notation to read when learning how to play the song yourself, but on the other hand it’s not exactly what’s in the recording, and the shift makes a big difference in how the riff feels. If you don’t believe me, try playing the eighth notes as straight as you can, and you’ll quickly see that the sudden lurch this extra time adds to Metallica’s performance is really significant. Every time I get up and headbang to “Master of Puppets,” I get a sudden burst of adrenaline when I hear this riff. This kind of visceral reaction entirely depends on the “inexact” timing Metallica uses, and the way it upsets with the movement of my headbanging body, for its effect.
If meter is defined as how you feel the beats of a measure, the last measure of this riff is certainly not 5/8, just because it feels very different. I’m not sure I could put another set of numbers into a time signature that would work better, but I think it’s important to recognize that this rhythm has a very different shape and physical implication than a straightforward measure of 5/8.
Hi:
Just discovered your website. Cool! Appreciate your analysis. I have posted a slightly different version of my comment here on my wordpress site and have included a link to your website. Hope that is OK. I’ve been thinking a lot about Master of Puppets album, and particularly the title track as I want to make a full arrangement for piano (some of the songs would be great as duets or 2 piano arrangements) (hopefully ones that aren’t lame sounding). Anyway, my take on the changing meter in the verse sections of “Master of Puppets” has been for a long time now been the following: 3 bars of 4/4 plus 1 bar of 3/4 – so really just a slightly unconventional 4 bar phrase. You can also hear it as a longer 2 part chunk as well: 8 + 7. I just picked up a published guitar transcription of the album at and I see the 5/8 written in there as you mentioned; it also is written with all this syncopation, which is weird because the drums clearly are hitting on downbeats and the vocals come right in on the 1st beat, so I don’t get the written transcriptions. Seems overly complicated, but I’m not hearing it as that complicated – it’s just uncompromisingly aggressive and propulsive which equals awesome! But I agree with you – that it doesn’t sound like 5/8 at all and I think your original assessment remains correct – it isn’t 5/8! What is suspicious is that in addition to the 5/8 bar is written all this syncopation. That is a clue to me that something is amiss here and the meters and written rhythms are being forced into a pattern that doesn’t really exist. It doesn’t sound anything like the way it looks. Anyway, I’m open to being convinced otherwise but I’ve been listening to the song with pretty focused intention for the last year or so and feel pretty comfortable with my assessment. Part of what gives the song its intensity and drive is not only the shortened 3-beat measure (as I’m hearing it) at the end of those verse phrases, it is also that it does not breathe at all – there is no “musical” space or relaxation between phrases like you hear in more traditional pop song (and classical) song performances. I think it’s worthwhile to point out that the typical liberties and relaxation of tempo (rubato?) in phrasing are rarely, if ever (that I know of thankfully) written into scores (except as a direction for a ritard) as a changing time signature or extra beats or syncopation. If they were it would be ridiculously difficult to learn and be really confusing. Regarding Master of Pupptes performance on the studio album, it sounds like sometimes there is a push to the downbeat. It’s on the verge of being out of control which is one of the really exciting things about it. But I wouldn’t put that on paper. I acknowledge that I could be incorrect; just seems like the 4+4+4+3 (or 8+7) pattern works consistently for those recurring verse sections of the song. Especially if you beat super strict “human” time (as opposed to a computer analysis) with absolutely no hesitations between the 8th notes. Anyway, that’s what I’ve been hearing. Open to further enlightenment on the subject. Best,
Just re-reading my 5 am, rather rambling “one thought’ comment above. If I may actually boil it down to one thought – it’s that the verse rhythm pattern is indeed 3 bars of 4/4 and 1 bar of 3/4. Today just ran into a Kirk Hammett YouTube video on master of puppet riffs which I believe leaves no doubts about this. https://youtu.be/GaaBtFdetTU
I’d be interested to know how your interpretation of the Audacity analysis might be affected by an assumption of 3/4 rather than 5/8.
Hi Steve, Thanks for reading my blog, and sorry I didn’t respond to your comment earlier! I had noticed earlier that when Metallica performs this song live, their timing of this riff is quite different from the album recording. Interestingly, in the demo you posted, it sounds like Kirk Hammett is playing another note that is not written in the last measure of the notated example I included in my post. In other words, it looks like he is playing E-G-A-E-G-A as even eighth notes, instead of E-G-A-G-A which appears in the “approved” transcription published on behalf of the band.
The average timing I measured for a quarter note in the music surrounding this measure is .29 seconds, and it’s a very stable average without much variation. A full measure of 3/4 would then be .87 seconds (.29 x 3), but in my measurements of the studio version on the album, this measure is consistently about .79 seconds. In other words, even if you analyze this measure as 3/4, it’s still consistently about a third of a quarter note out-of-time on the album.
I measured timings on the version you posted, and in that performance Kirk Hammett actually plays slightly slower than 3/4 (the quarter note in that recording averages .30, and the measure we’ve been talking about takes 1.00 seconds). Definitely a big difference from the timing on the album, which is interesting. Which version do you prefer, the studio version on the album, or any live recording?
Hi,
Cool analysis. The mathematics you’ve worked out kinda answer the question themselves – if, using a bar of 5/8 as a starting point, we need to account for an extra .05 of “clock” time, which using your maths is a third of an eighth note, we’ll never get an accurate transcription of the “true” time of this bar because we simply can’t get a measurement in music which will break our bar length down into thirds of eighth notes. Using your calculations to take 0.8 as the length of this bar, if you get down to a bar of 43/64 (which at this BPM is waaaay past anything you could conceivably count or apply in real life) you’re pretty much there (0.80625), but I think, given that what’s actually going on here is just a well-oiled band feeling an arbitrary pause in the music, you might as well stop the buck at 5/8 or at the very most 11/16, and perhaps notate the brief pause another way.
Hi Stephen,
thank you for your detailed analysis. You saved me some Audacity work!
I am transcribing this song at present and am not happy with using a 3/4 bar, as it is too long. On the other hand, a 5/8 bar is too short.
The math:
If a quarter note is 0.29 secs, then dividing by 4 gives us a sixteenth note rate of 0.0725. Multiply this number by 10 (10 sixteenths in 5/8) and we have a total of 0.725. Quite a bit short of the 0.7975, the average duration of the measure in question. So, 5/8 is too short.
Alternatively, using a 3/4 measure (12 sixteenths) would give us a total length of 0.87. Too long.
However, if we multiply the sixteenth note rate by 11 we get 0.7975; a hair’s breadth from the 0.79 average. I have checked this theory with a metronome and believe this to be the most accurate representation of this rhythm.
The sixteenth notes in the 11/16 bar are grouped 2+5+4.
There is something special about how Metallica played this figure. I believe it’s worth notating it accurately!
All the best,
I agree this is the right way. Just quickly tabbed all the suggestions from here into guitar pro, and that one is by far the closest. Also had an idea of 21/32, where the extra 0.04 seconds would be a break, but that one made the break too short.
I’d tab it 0-3/5-r-0-3/5 with the r being a 16th break and the rest is even 8th notes.
Thanks for that, Henri. I also used computer playback to compare the various approaches. You can see and hear them here:
http://www.onlinedrummer.com/drum-beats/master-of-puppets-metallica-verse/
Cheers
Hi Stephen — Just stumbled on this great post. I’ve been thinking about this riff, too. At first, I was leaning toward the “stretched 5/8” interpretation, but I think there’s stronger evidence for a “shortened 3/4.” The YouTube link that Steve shared gives a pretty clear 3/4 (though slightly long, as you point out), which makes me think Hammett always thought of the figure as having six eighth-notes. This would mean that, in the album version, the second shaded column in your chart is accelerated, which makes sense if we check out the 3+3+2 figure at :25, :30, :39, and :48 — where the middle 3 is consistently short by ~50-70ms. In other words, there’s a precedent for the “rush” approach there. Finally, at 5:46 and 5:59, the guitar solo seems to articulate three beats when playing over the spot in question. A fun puzzle!
Hi Fernando,
Thanks for your encouraging comments! In re-watching Hammett’s video lesson, I think it’s absolutely clear that he plays the figure in 3/4, especially when he plays a slowed-down version.
I also agree with what you say about the 3+3+2 rhythm being rushed. I think Metallica rushes these kinds of rhythms a lot, at least in their earlier albums. I’ll be putting up a new post about that soon!
For those who don’t know, Fernando Benadon is also a music theorist! I read one or two of your articles for a class in my first quarter of graduate school, so it’s an honor to have you visit my blog and share your opinion.
Best,
–Stephen
Hey there! I believe diving into the track at this level is at the point of rubato dealing.
I think the odd times are actually pretty straight forward.
People often try to say it’s a sudden 5/8 or perhaps a wierd 11/16 when it’s much more comfortable to count it in the tempo the band is thinking in instead of frantic changes.
I believe they count in eigth notes – so they don’t even count it as a slow 4/4 and then bam super fast 11/16. it would make much more sense to count it evenly. But how?
Well simple take the 4/4 precedented to the “11/16” which would make some sort of 15/16.
I believe it’s actually 15/8. Just count it 8/8 + 7/8.
I think it really is that simple. It’s like how people thought that OMG sufjen stevens’ UFO Sighting Near … is 18/8 or something, and kept on counting and counting untill they realized it’s some sort of rubato. Later in a documentry he said that the song has no time signature and he switches when he feels like on the album. In the end it seems that live with the band they just play 9/8 and honestly, it’s a very simple song. So we gotta stop get so frantic and just enjoy the music to the extent that is re-playable in an exact manner.
I think the most obvious solution to his is that the first note in the “5/8” bar is actually a 16th making the two snare hits on the e of one and the 2 of a 2/4 bar
This would make the interval of the two snare hits a dotted 8th and account for the “30% longer 8th note”.
So the last two bars would just be counted “1 and 2 and 3 and 4 and 1 E 2” – again with that final E and 2 being the snare hits
Another reason I lean this direction is that if the 5/8 bar were correct, we should expect to feel the down beat go upside down when tapping quarter notes, but it doesn’t. If you tap quarters along with the tune, the down beat stays in the same spot
I’ve yet to come across anyone that is describing what I’m actually hearing.
I’d write it simply 3 bars of 4/4 and a bar of 2/4 – the final bar counted as 1 e 2 – the two snare hits landing on the e of 1 and the 2 – think of the rhythm of “ba dum bum” when someone tells a joke – that final bar is the exact same rhythm.
So written all the way out it’d be:
1 and 2 and 3 and 4 and
1 and 2 and 3 and 4 and
1 and 2 and 3 and 4 and
1 E 2 Repeat
This is a much simpler solution to the problem and actually corresponds with anomalous .07 second beat (obviously a 16th note).
Also, if you tap quarter notes to the actual recording, you’ll find the down beat never changes, but if there were a 5/8 bar (or some other variation), we would expect the down beat to switch – but when you tap quarters to the recording, it clearly does not switch.
I’d love to have someone give that a try and weigh with their thoughts.
Hi Micah! First, thanks for reading my blog! I think what you’re proposing makes sense as a way to count it. You’re right about how one would expect the beat to “switch” if there were a true 5/8 measure, and my ears agree with you that it doesn’t feel like this kind of switch happens in this passage.
From a clock time perspective I’m not sure your counting any closer than the other options: if you set up a metronome to the tempo of the 4/4 measures, it will definitely be off after that “different” fourth measure (unlike some other places where Metallica uses a 2/4 measure at the end of a 4/4 phrase and stays in tempo). But there’s something about the way of counting you describe, as two beats with a syncopated accent, that makes the extra time feel fairly natural. That’s a very interesting observation and I will have to think about it more! 🙂
I have a pdf of master of puppets . It shows 11/16 for the last part. I don’t get time signatures that much. It is 3x 4/4 and 1x 11/16 bars. If I understand the problem correctly ? : notation shows that we must play this 4 bars 2 times. Then ,after playin this 4 bars one time , we have to add 16th notes’ time to first beat when we re-play for second time. After playing this 4 bars 2 times, the actual beat shifts 8th notes’ time. It sounds too complex to me ,something must be wrong.I will ask this rhytym to a music teacher who has PHD and a really good Gypsy Jazz player next week I’m not native english speaker btw…
I’m currently re-transcribing this in Guitar Pro 7 (I made a GP5 tab of it in 2013 and I’ve never been happy with it, particularly the bar in question). I’m glad to see I’m not the only one who doesn’t like the standard transcription! There’s definitely something going on with that bar.
I’m not a music expert or anything, but I’ve been experimenting with the isolated stems, adding a clicktrack to the drums and seeing what happens when it hits that bar. It sounds like the clicktrack remains in time when the subsequent bar comes in, so it must be of uniform length even if the notes within it aren’t, and the overall tempo must be the same too (I experimented with adding subtle tempo changes in that one bar to see if I could make it fall into place). After some jiggery-pokery, I came up with something that “feels” right, to me at least, but which looks quite messy in the tab. I’m curious if it will yield an accurate result when you apply your math wizardry to it (I’m afraid that’s beyond my capabilities).
In a 5/8 bar, add the following:
A standard 8th low-E string note, followed by a quintuplet comprising of the two 8th G>A slides and the 8th rest between them. Here’s a pic for clarity:
https://i.imgur.com/nXBj89t.png
It at the very least feels better than the usual interpretation. I think there could be something to the idea of weird n-tuplets, rather than a time signature issue.
I’ve also discovered that those sliding chords are not power chords, but octave chords, like this:
(5 / 7)
x
(3 / 5)
Occasionally the A-string is sounded, presumably by accident, so you get that C note coming in from time to time. I checked the Kirk Hammett lesson on YouTube and he plays the octave chords there, although I don’t know if they bother with it when playing live.
Thanks for your investigation of this, there’s some really interesting and valuable information here! A lot of it went over my head of course, but it’s fascinating nonetheless.
Cheers!
One of the only posts I’ve seen anywhere that gets the octaves in the mystery bar right. That’s 100% what they’re doing. Noticed it years ago in that same Kirk video. (which is unintentionally hilarious because he has the completely wrong official tab book propped up on the amp behind him)
Whoa I never noticed that! I went to check, and you’re right, you can even see Kirk’s pinky finger on the 3rd string in this live video if you pause on the right frame. Cool!
https://youtu.be/cJOSvvdy27I?t=106
Those tab books are probably single-handedly responsible for 40 years of guitarists not knowing that those are octave chords 🤣
I just finished retabbing my ancient Puppets tab if you’re interested (which now includes the octave chords, yay!)
Guitar Pro 8 format:
http://pmtabs.net/tabs/gp/Metallica%20-%20Master%20of%20Puppets.gp
Text version also available:
http://pmtabs.net/tabs/guitar/Metallica%20-%20Master%20of%20Puppets.txt
Just a heads up: I moved to a new host, so those links above no longer work. The new links are on https://pmtabs.net/ if you don’t trust the following direct links:
Guitar Pro 8
https://pmtabs.net/wp-content/uploads/Metallica-Master-of-Puppets.zip (my new host doesn’t like .gp files for some reason)
Text/ASCII:
https://pmtabs.net/wp-content/uploads/Metallica-Master-of-Puppets-guitar.txt
And here’s the bass as well:
https://pmtabs.net/wp-content/uploads/Metallica-Master-of-Puppets-bass.txt
Hey, very interesting topic here!
I think this strange bar could come from the inspirations Metallica (and metal bands in general) are taking from Classical Music. In classical music, it is usual to see an orchestra play slower or faster for just a small part of a tune, due to how they interprate the score.
I think we could say that something similar happening here with this strange bar in the verse.
Anyway, thanks for the interesting article !
After doing further research, I’ve now come to a rather boring conclusion: it’s a 3/4 bar with six 8th notes (one of which is the rest after the first slide).
Your analysis says you stopped at the first verse, but if you continue to the guitar solo (which uses the same verse riff as a backing) you will run into a problem if you try to transcribe it: the guitar solo is playing as if it’s a 3/4 bar with 8th note triplets. If you try to tab this in a 5/8 bar, you will find it impossible to make the notes fit without it sounding markedly off before the next 4/4 bar kicks in. In a 3/4 bar, it fits like a glove.
One way to illustrate this is using Audacity to take the isolated guitar track (which is available in a pack of stems ripped from Guitar Hero, and during the guitar solo the rhythm guitar is shifted to a different stem), cut out everything leading up to the guitar solo, then copy the first two 4/4 bars of the solo and paste it repeatedly beneath the main solo track in a new track. Although the notes being played won’t sync when you hit the 3/4 bar, the rhythm/duration of them will. This showed me that the duration of the notes don’t take on a different form during the bar in question, they stay the same as they are in the other bars. If that’s the case, the bar simply has to be 3/4.
Likewise, if you do the same thing with the isolated drum track, that is take the verse riff, paste the first two bars of 4/4 under it repeatedly, you will find that the 5/8 form just doesn’t work if you keep count, but 3/4 does. The duration of the two snare hits in the 3/4 bar also conforms to 8th-note layout (in that they are a dotted quarter note followed by a quarter note, respectively), rather than there being micro-delayed/lengthened notes or rests.
I did the same thing with the bass, which is probably the one instrument that makes it obvious what the timing is, because although the rest of the band has that distinctive rest in the middle of the bar, the bass often plays a “throwaway” ghost note during the rest which is an 8th note, and the rest of the notes the bass plays are also 8th notes.
Along with all of the above, the main way I came to this firm conclusion was by tabbing this section in Guitar Pro 7 (at 211 BPM), exporting the rhythm guitar track’s audio, and then pairing it up with the isolated rhythm guitar stem. I made sure to get the first 4/4 bar to sync up, which is quite easy given that it’s 8 evenly-spaced low-E notes, so if there was any desynchronisation or tempo misalignments it would be noticeable. To aid in detecting such non-synced parts, I panned the real guitar to the far left and the tabbed audio to the far right. You could also bump the tabbed version up an octave for added clarity if needed.
The first thing I learned from this is that it’s not a 5/8 bar, because with a 5/8 bar the recording lags behind the tabbed version by a single 8th note after the bar in question. If the tabbed version is in 3/4, the subsequent bars sync up properly and maintain their synchronisation for the rest of the verse. This is also what led me to figure out why the guitar solo wouldn’t fit that bar no matter what I did to it while working in 5/8.
The second thing I learned is that each of the chords/notes in this 3/4 bar are definitely 8th notes, because they sync up perfectly with the tabbed version (in which I used 8th notes throughout).
This probably sounds dubious given the amount of work you put into your analysis, but I think the micro-differences you’ve noticed might just be from human playing and the ways in which audio that’s been slowed down sufficiently will make it sound like certain notes are of differing lengths, rather than intentional convention-breaking weirdness. I run into this issue when I transcribe fast guitar solos (especially Kirk’s, which are often quite frantic); the extremely-slowed down version makes it sound like the notes are staggered and non-uniform, but speed it back up to even just 50% tempo and these little flickers of mistiming disappear. I’ve found myself tabbing things so precisely at 20% speeds that I have to use bizarre note groupings and such, when in fact I didn’t need all that and just needed to pull back a little and loosen my grip on it.
I can supply you with the aforementioned stems if you need them 👍
Whoa I am impressed at the work you put into this! I took a listen and you are absolutely right about how during the guitar solo, it is pretty clear that this measure is in 3/4 in that part of the song. The same riff in the Verse sure doesn’t ever sound like 3/4 to me though.
There’s always room to speculate that maybe Metallica thinks of this riff as clearly in a particular time signature, but just plays it rushed in some parts of the song. Certainly, Metallica tends to rush some other phrase-ending “turnaround” rhythms like this one; something similar happens in the Bridge of “Kill ‘Em All,” where there’s a riff that is sometimes transcribed in 7/8, but I think is really a 4/4 rhythm that is rushed. http://metalintheory.com/metallica-kill-em-all/
But I think I would take it further, and suggest that Metallica isn’t thinking of it in 3/4 and sometimes rushing it; and they also aren’t trying to intentionally do some weird micro-timing voodoo subconscious trick. I think the simplest explanation is that they aren’t counting it at all, just playing by feel. There’s a great passage in Mick Wall’s biography of Metallica that makes it sound like Lars Ulrich didn’t know anything about conventional ways of thinking about counting and meter as late as 1984. This makes me think that maybe it’s misrepresenting things to try to say whether it’s in 3/4 or 5/8 or whatever, because maybe the band doesn’t think of it in terms of beats or time signatures at all. That would certainly explain why the same rhythm seems to be played slightly differently in different parts of the song, and during the Verse it seems to be kind of ambiguously somewhere in between 5/8 and 3/4.
“Lars’ difficulties on the drums were more problematic. ‘I thought he was absolutely useless,’ Flemming says now. ‘I remember the very first thing I asked when he started playing was: “Does everything start on an upbeat?” and he went, “What’s an upbeat?” Holy shit! The thing is that Lars is an innovative person, so his whole drumming had been based on drum fills. That was his thing. All the ones and twos in-between, he never took notice of that. He didn’t really think about what was going on between the drum fills. I still think he’s a great drummer in his own right ‘cos I think he does some things that are absolutely amazing. But me and the guy who was his drum roadie, another guy called Flemming [Larsen] who at that time was [also] playing drums in a Danish metal band called Artillery, we started telling him about [beats]. That they have to be an equal length of time between that hit, that hit and that hit and you have to be able to count to four before you come in again … [Then he could play] a really good fill that nobody else had thought of doing at that time.’ He pauses then adds, ‘I can’t imagine what they must have been like live at that time. He was speeding up and down in tempos a lot [playing] more the way he felt the songs SHOULD be.'”
Mick Wall (2011) pp. 170-171
Hi there,
just wanted to share my thoughts on the topic from a different (mathematical) perspective.
It struck me as odd that the suggested time signatures for the last bar, such as 5/8, 11/16 and 21/32, are closest approximations of the fraction 2/3 with a power of 2 in the denominator. So I thought, perhaps the idea behind the additional pause is to make the duration of the bar roughly 2/3 of the 4/4 bar? Sure, that’s an unusual fraction to have in music, but the riff does seem to work in a very unusual way.
After doing some calculations on your timings, I got that the duration of the last bar in the recording is 68.1% of the 4/4 bar. In reality this value might be slightly different because the timings in the table were rounded to 2 decimal places, but still, 68.1% is much closer to 2/3 (~1.43% diff.) than a 5/8 to 2/3 (~4.17% diff.) and slightly closer than a 11/16 to 2/3 (~2.08% diff.), which could explain why the riff sounds off even when played in complex time signatures. Interestingly, 21/32 would actually be a better approximation (~1.04% diff.) of 2/3 than what Metallica played according to the measurements.
Cheers.