The Internet Juggling Database


endeesjafrcaitherudanlel

Understanding Siteswap Juggling Patterns - A guide for the perplexed

Greg Phillips - 2nd December, 2001.

greg.phillips@juggler.net

This guide started life as a handout for a workshop I taught at the 2001 Kingston Juggling Festival. My aim was to produce a concise summary rather than a complete tutorial; I hope I've succeeded. For your printing convenience the guide is also available as a one-page Adobe Acrobat file in either A4 or letter format. Thanks to Michael Ferguson and Todd Strong for comments on earlier drafts. — Greg

Core rules (for all patterns)

C1 We imagine a metronome ticking at some constant rate. Each tick is called a "beat."
C2 We indicate each thrown object by a number that tells us how many beats later that object must be back in a hand and ready to re-throw.
C3 We indicate a sequence of throws by a string of numbers (plus punctuation and the letter x). We imagine these strings repeat indefinitely, making 531 equivalent to ...531531531... So are 315 and 153 — can you see why?
C4 The sum of all throw numbers in a siteswap string divided by the number of throws in the string gives the number of objects in the pattern. This must always be a whole number. 531 is a three-ball pattern, since (5+3+1)/3 equals three; however, 532 cannot be juggled since (5+3+2)/3 equals three and a third.
C5 Throws must balance catches. By rule C4, 432 appears to be a three ball pattern — but it can't be juggled since the 4, 3 and 2 would all need to be re-thrown at the same time and the pattern says they aren't. By shuffling the numbers we find 423, which can be juggled — try it!

Visualizing the numbers

At any given beat rate, higher siteswap numbers mean higher throws. A faster beat means lower throws for all numbers. You can understand the numbers as meaning roughly:

  • 0 — an empty hand
  • 1 — a quick pass across, as in a shower
  • 2 — a quick "throw to the same hand" (normally just a hold)
  • 3 — a three-ball cascade throw
  • 4 — a four-ball fountain throw
  • 5 — a five-ball cascade throw... and so on up to 9, then
  • a — a ten-ball fountain throw (write a instead of 10 so we don't confuse "ten" with "one-zero"; read a as "ten")
  • b — an eleven-ball cascade throw... and so on
Asynchronous siteswap (alternating throws)

Many juggling patterns are based on alternating right and left-hand throws. We describe these using asynchronous siteswap.

A1 The right and left hands throw on alternate beats.

Rules C2 and A1 together require that odd-numbered throws end up in the opposite hand, while even numbers stay in the same hand. Here are a few asynchronous siteswap examples:

  • 3 — a three-ball cascade
  • 42 — two balls juggled in one hand, a held ball in the other
  • 330 — a three-ball cascade with a hole (two balls)
  • 51 — a three-ball shower
  • 53 — a four-ball half-shower
  • 64 — three in one hand with one hand, two in one hand with the other (each at its regular height — this is hard)
  • 73 — a five-ball half shower (the harder one!)
  • 615150 — a three-ball shower with a "leak"
Synchronous siteswap (simultaneous throws)

Some juggling patterns involve both hands throwing at the same time. We describe these using synchronous siteswap.

S1 The right and left hands throw at the same time (which counts as two throws) on every second beat. We group the simultaneous throws in parentheses and separate the hands with commas.
S2 We indicate throws that cross from hand to hand with an x (for 'xing').

Rules C2 and S1 taken together demand that there be only even numbers in valid synchronous siteswaps. Can you see why? The x notation of rule S2 is required to distinguish patterns like (4,4) (synchronous fountain) from (4x,4x) (synchronous crossing). Unlike a 2, a 2x must always be thrown since it changes hands. Here are a few examples:

  • (4,4)(4,0) — three-ball columns (a.k.a. Two Up, One Up)
  • (4,2) — appears identical to 42 — The Fake, The Yo-yo, etc.
  • (4,2x)(2x,4) — The Box
  • (6x,4x) — the easier five-ball half shower
  • (6,6)(6,6)(6,0) — Enrico Rastelli juggling five plates
Multiplex throws

In both alternating and synchronous patterns we can throw multiple objects from the same hand at the same time.

M1 Multiple objects thrown from the same hand count as one throw. We enclose their numbers in square brackets.

Applying C4 and M1 to the siteswap [33], we sum the throw numbers (3+3) and divide by the number of throws (one) to give six — so [33] is a six object pattern. Note that in multiplex patterns twos inside square brackets are almost always thrown rather than held. A few multiplex examples:

  • [33] — six balls juggled in groups of two as a three-cascade
  • 4[43]1 — a pleasant four ball variant of 441 , itself quite fun
  • [32] — five-ball splits
  • ([44],[44])(4,0) — Four Up, One Up
A few siteswaps worth learning

Here are a few reasonably easy but fun siteswap patterns:

501, 423, 441, 4413, 531, 5313, 534, 55244, 561, 633 (easier with bounced sixes), [33]33, [33][33]3, [33], [43]1421, 4[43]1, [32], (4,4)(4x,4x)

Training for five: 50500, 52512, 55500, 50505, 552, 55550

What siteswap doesn't show

Siteswap notation doesn't show us how a throw is made; for example Mills Mess has a siteswap of 3, which completely ignores the sinuous arm crossing that makes it beautiful and fun to juggle. On the other hand it is useful to know that both Burke's Barrage and Windshield Wipers are juggled as 423, even if that doesn't tell the whole story. Happy juggling!


post a new message
29th May 2002
Kevin: I understand your comme...
Kevin: I understand your comments, but I think you've missed the point. The aim of this page is not to provide "definitive, correct" siteswaps for any particular patterns (that would be hopeless anyway) but to provide a few simple rules and illustrations to help newcomers understand how siteswap works. Any siteswap model is just that --- a model --- and models can be more or less accurate but still useful. Besides which, the way you've written your siteswaps is *very* non-standard; I don't think there's a juggling simulator on the planet that would accept any of what you wrote.
2nd Jun 2008
Re: Kevin: I understand your comme...
I read the article myself but it's hard to form your own patterns using numbers, and the , [ ( or x. I mean, I think the purpose of this notation is just to help people to get a better grasp at their patterns. Nothing more really. It's not like music notation where you can just read it and know what it is. For this notation, you have to do some trial and error I guess.
12th Sep 2004
what about pirouettes? (spelli...
what about pirouettes? (spellin)
25th Dec 2002
This is simple to understand i...
This is simple to understand instructions. I see you have an eye for excellence. Thanks
23rd Oct 2002
aha now i know my knob from my...
aha now i know my knob from my butt so to speak
(no one said anything about club jokes ;-)
17th Apr 2002
Oh yeah, and 4[43]1 is a bit s...
Oh yeah, and 4[43]1 is a bit suspect too, because the uniplexed (?) 4 has to be caught on the same beat (and in the same hand) as the duplexed 3. It's not impossible, but it's bloody hard. The pattern you're thinking of might be:
([5s 4x],0)-(0,1)-(4,2)-(0,0)-(0,[5s 4x])-(1,0)-(2,4)-(0,0)
which is a lot easier even if the siteswap looks crap. It still feels like 441 but it's a 4-beat pattern, not 3. Hey ho.
16th Apr 2002
Disagree with [32] for 5-ball ...
Disagree with [32] for 5-ball splits. Assuming you actually throw the 2, you end up having to catch this and the 3 from the opposite hand at the same time, which is not what most people mean by 5-ball splits. Throw heights of 3 and 2 also feel far too 'low'. I think [32] is actually a 3-ball cascade, holding a spare ball in each hand (the 2's - not thrown).

For 5-ball splits, I get:
([5 4],0)-(0,1s)-(0,[5 4])-(1s,0) for a 'high-low' duplex, and:
([4x 4],2)-(0,0)-(2,[4x 4])-(0,0) for an 'even height' duplex, where:
a) every beat is shown (i.e. not every second beat);
b) 1s (self 1) signifies a 1-beat hold;
c) 2 signifies a 2-beat hold, and;
d) 2's are only shown on the beat the hold is 'thrown', if this makes sense (i.e. the (0,0) in the second pattern is correct even though a ball is being held at the time, since this hold started on the previous beat).
18th Dec 2001
this is a nice piece of work e...
this is a nice piece of work explaining siteswap.
hope you read my book about it when it comes out.