Skip to content

← Math Further 6–8

How Does an Abacus Think?

Topic 39 · Grades 6–8 · The abacus and place value

What this builds on from class
In this unit 9 sections
  1. One rod, ten digits
  2. Friends of five
  3. Past ten: friends of ten
  4. What’s going on
  5. Why one bead above and four below?
  6. An abacus in your head
  7. Where this is used
  8. Questions
  9. Where to go next

Your cousin Anisha is eight and goes to abacus class on Saturdays. You read out a sum to test her: “4 + 3 + 8 + 9.” Her right hand flicks at the air in front of her, as if it were moving beads that aren’t there, and before you have worked it out yourself she says “24”.

Her abacus at the class is a Japanese soroban. Each rod has just five beads: one above a bar, worth 5 when it is pushed down to touch the bar, and four below, each worth 1 when pushed up to touch it. Beads away from the bar count for nothing.

Here is the first puzzle she had to solve. The ones rod shows 4: all four one-beads are up against the bar. Now add 3. There is no one-bead left to push up. What do you do?

Guess first: on a rod showing 4, how do you add 3?

Pick an answer to see why.

Try it

Add on an abacus

Tap a bead to move it (it pushes the beads in its way along with it). Each tap is one move, and every move is written down as what it adds or takes away. Do 4 + 3 in as few moves as you can, then the other sums.

0hundreds0tens4ones
On the abacus
4
Goal
4 + 3 = 7
Moves
0

Your moves: —

The abacus shows 4. Add 3 by tapping beads, in as few moves as you can.

One rod, ten digits

Reading the beads

A rod shows one digit, and you read it by counting only the beads that touch the bar. The five-bead down means 5; each one-bead up adds 1. So five beads are enough for all ten digits.

Five-bead up: 0 to 400 + 010 + 120 + 230 + 340 + 4Five-bead down on the bar: 5 to 955 + 065 + 175 + 285 + 395 + 4
Figure 39.1 The ten digits on a soroban rod. In the top row the five-bead is up, away from the bar, and the rods show 0 to 4. In the bottom row it is down on the bar, and the same one-beads now show 5 to 9.

Each rod of the abacus is one place: on the right are the ones, then the tens, the hundreds and so on to the left, exactly as you write a number. The rods are place value you can hold, from Grade 6. A rod with no bead touching the bar is a 0, an empty place. A soroban usually has 13 rods or more, so it can hold numbers of 13 digits.

A soroban’s four rods show, from left to right: five-bead down and one one-bead up; nothing touching the bar; one one-bead up; five-bead down and three up. What number is it?

Pick an answer to see why.

Seeing, not counting

Why 5 and 1, and not ten beads worth 1 each? Try reading 8 as eight beads in a row: you would have to count them. On the soroban, 8 is “the five, and three more”, and three beads can be seen at a glance. Psychologists who study the abacus point out that its beads come in groups of four or five, about as many things as the mind can keep track of at one look. So a rod is read in an instant, and a row of rods almost as fast.

Friends of five

On the soroban you never stop to work out the answer and then set it. You move beads by a few fixed rules, and the answer appears by itself. The first rules are for adding when the one-beads run out.

You saw one: on a rod showing 4, add 3 by adding 5 and taking away 2. That works because +3 = +5 − 2. 3 and 2 make 5 together; we’ll call them friends of five, and in maths they are called complements: two numbers are complements of 5 if they add up to 5.

To add push the five-bead down and take away because
1 4 one-beads +1 = +5 − 4
2 3 one-beads +2 = +5 − 3
3 2 one-beads +3 = +5 − 2
4 1 one-bead +4 = +5 − 1

You only need this rule when there aren’t enough one-beads down to push up. 1 + 3 needs no friend: three of the four one-beads are still free. The widget above counts your moves: 1 + 3 takes one move, 4 + 3 two.

A rod shows 3. What do you do to add 4?

Pick an answer to see why.

Past ten: friends of ten

Now the rod itself runs out. A rod showing 8 has its five-bead down and three one-beads up. Add 4, and the answer, 12, needs a second rod. So you use the next bead up: one one-bead on the tens rod is worth 10. Push it up (+10) and take away 6 on the ones rod (−6). Altogether that is +4, because +4 = +10 − 6. 4 and 6 are friends of ten, complements of 10.

To take away 6 from the 8 on the ones rod, you push the five-bead up (−5) and one one-bead down (−1). So the whole of 8 + 4 is three moves: +10 − 5 − 1.

Try it

Add past ten

Do each sum in as few moves as you can. Then look at your moves: which friend of ten did you use each time?

0hundreds0tens8ones
On the abacus
8
Goal
8 + 4 = 12
Moves
0

Your moves: —

The abacus shows 8. Add 4 by tapping beads, in as few moves as you can.

Hint

The friends of ten are 1 and 9, 2 and 8, 3 and 7, 4 and 6, and 5 and 5. To add a number when the rod hasn’t room, add 10 on the next rod and take its friend away here. Sometimes the taking away needs a friend of five as well. 7 + 6: push one up on the tens rod (+10), then take 4 away from the 7 on the ones rod. But the 7 has only two one-beads up, so −4 is done as −5 + 1: the five-bead goes up and one more one-bead goes up. The ones rod now shows 3, and the abacus 13.

And when the next rod is full too, the carry runs along. 99 + 1: the ones rod has no room, so +1 = +10 − 9, but the tens rod shows 9 and has no room either, so that +10 becomes +100 − 90. Every rod of 99 is cleared and one bead goes up on the hundreds rod: 100. It is the same carry you do on paper, when 9 + 1 makes 0 and you carry 1.

What’s going on

Place value you can hold

Every rule above comes down to place value. A bead on the tens rod is worth ten beads on the ones rod, because each place is worth ten times the place on its right. So pushing up one bead on the tens rod and taking ten away on the ones rod changes nothing; and adding 4 can be done as adding ten and taking away six. The abacus never “knows” that 8 + 4 is 12. It follows the rules, and the beads end up showing 12.

This is the idea behind the topic’s question. An abacus doesn’t think: you do. What it does is hold every number so that a few small rules, each needing no sums at all, do the adding.

Two counters on every rod

Look again at the ten digits in Figure 39.1. Each digit is made of two parts: the five-bead (down or up, two choices) and the one-beads (0 to 4 of them, five choices).

digit=5×(five-bead: 0 or 1)+(one-beads: 0 to 4) \text{digit} = 5 \times (\text{five-bead: 0 or 1}) + (\text{one-beads: 0 to 4})

A code built from a part with two choices and a part with five is called bi-quinary (bi, two, and quinary, five). Every digit has exactly one pattern of beads, and there are 2 × 5 = 10 patterns, just enough.

In Grade 8 you write numbers in base 5, where each place is worth five times the one on its right. Look at a single rod with that in mind: the five-bead is the fives place and the one-beads are the ones place. So each rod writes its digit in base 5! 7 is a five and two ones, and in base 5, 7 is 12₅. 9 is 14₅, five-bead down and four up. Every rod is a tiny base-5 number, and the rods together make a base-10 number.

In base 5, what does the rod showing 8 write?

Pick an answer to see why.

Why the friends always work

Does one of the rules always fit, whatever the rod shows and whatever you add? Here is the argument, for adding a digit from 1 to 4 to a rod.

  • If there are enough one-beads down, push them up. Done.
  • If not, most of the one-beads are already up. Say you are adding 3 and fewer than 3 are free: then at least 2 are up. If the five-bead is still up, push it down and take away the friend, 2. At least 2 one-beads are up, so there are always 2 to take away.
  • If the five-bead is down as well, the rod shows at least 5 + 2 = 7, and 7 + 3 is at least 10: too big for one rod. Then add 10 on the next rod and take away 7 here, which works because the rod shows at least 7. Adding 10 is adding 1 on the next rod, so the same rules apply there, and if that rod is full too, the carry runs on, as in 99 + 1.

The same thinking works for every digit you add. For 5 to 9 the five-bead goes down with the extra one-beads if they fit; if they don’t, or the five-bead is down already, the rod plus what you add passes 10, and a friend of ten does it. A computer checked the rules bead by bead for every sum of two numbers from 0 to 999: 1,000,000 sums, every one right. That is why abacus addition never needs a table of sums: just the friends of five and of ten.

Why one bead above and four below?

Five beads is the fewest that can show 0 to 9 with beads worth 1 and 5. With only three one-beads you couldn’t show 4, and with no five-bead you would need nine one-beads to reach 9. But other abacuses chose differently.

suanpan (China)2 + 5 beadsshows 0 to 15soroban (Japan)1 + 4 beadsshows 0 to 9schoty (Russia): 10 beads, all worth 1
Figure 39.2 One rod of three abacuses, each showing 7. The Chinese suanpan has two beads worth 5 above the bar and five worth 1 below; the soroban has one and four; the Russian schoty has ten beads on a wire, all worth 1, with the middle two darker to make the beads easier to read.

The Chinese suanpan has seven beads a rod, two fives and five ones, so a rod can show up to 2 × 5 + 5 = 15. That looks like a waste for digits that stop at 9, but the extra beads were useful. In China’s old weights, 1 jin was 16 liang, so sums of weight needed places worth sixteen times the place on their right, base 16, and a rod that holds 0 to 15 is exactly what base 16 needs. The extra beads also hold a carry for a moment while you work. Japan took the Chinese abacus and dropped the beads digits don’t need: first one five, then one one.

The Russian schoty goes the other way: ten beads on each wire, each worth 1, with no five-bead at all. It needs no friends of five, but more beads to slide and to count, which is why its 5th and 6th beads are colored differently: they make the beads easier to read without counting.

An abacus in your head

Once they are quick on a real abacus, students practice mental abacus (in Japanese, anzan, “blind calculation”): they picture the soroban and move the beads in their mind. The rules are the same; only the beads are imagined. The hands seem to be part of it, as Anisha’s were: researchers found that keeping trained users’ hands busy upsets their sums more than making them talk.

How much abacus can a mind hold? Try it.

Try it

How many rods can you hold?

Press Show. The abacus appears for a second and a half and is then hidden: pick the number it showed. Every right answer adds a rod; every wrong one takes a rod away. How many rods can you hold?

Press Show3 rods
Rods next time
3
Most rods right
—

Press Show. The abacus is shown for a second and a half, then hidden. Which number was it?

How far did you get? Researchers who tested children trained in mental abacus in India found that even they seem to hold only about three or four rods at a time. Each rod is held in the mind as one picture, a single “thing” with its beads inside it, so the imagined abacus is a row of a few rod-pictures, and long numbers are worked a few rods at a time. The same research found that trained users can calculate while they talk, which badly upsets ordinary mental arithmetic: the imagined abacus seems to be kept as a picture, not as words.

Where this is used

Questions

a.★

Say what each rod shows (the five-bead down or not, how many one-beads up) for 63, 509 and 8,848 (Everest’s height in meters, leaving out the 0.86).

Answer

Which pair are friends of ten?

Pick an answer to see why.

b.★★

Write the moves, as + and − of 1, 5 and 10, for each sum done on the ones rod (and the tens rod where needed).

  • a.2 + 4
  • b.6 + 3
  • c.9 + 6
  • d.5 + 8
Answer

How many separate moves (a tap of the five-bead or of a group of one-beads) does 999 + 1 take on a soroban, at the fewest?

Pick an answer to see why.

c.★★

How many beads touch the bar when a four-rod soroban shows 2083? Which four-digit number needs the most beads on the bar, and how many?

Answer
d.★★

Write each digit as its rod writes it in base 5: 4, 5, 6, 9. Then explain why no soroban rod ever shows a base-5 number bigger than 14₅.

Answer
e.★★★

A suanpan rod has two beads worth 5 and five worth 1.

  • a.What is the biggest number one rod can show?
  • b.In how many different ways can a suanpan rod show 5?
  • c.Why did this make the suanpan useful for weights in jin and liang, when 1 jin was 16 liang?
Answer
f.★★★

Invent an abacus for base 5 (Grade 8): how many beads would each rod need, and what would each be worth? Show 38 on it. Then say how its rule for adding past the end of a rod would read.

Hint
Answer

Where to go next

Where these facts come from
  • The soroban (one bead above, four below, a bar, an odd number of rods, usually 13 or more), bi-quinary digits, addition by complements, anzan, the Cranmer abacus and toggle abacus for blind users, the bead history (two and five, one bead removed around 1850, Irie Garyū in 1891, the four-bead kind reintroduced in 1930 and popular in the 1940s; arrival in Japan in the 14th century by most historians, about 1600 by some), the 1946 contest’s rounds and the Nippon Times and Stars and Stripes quotes: Wikipedia, “Soroban”, read October 2026.
  • The Chinese abacus reaching Japan “a little past the middle of the fifteenth century”, the five-bead and four-bead soroban together towards the end of the 19th century, the four-bead soroban in the national arithmetic textbooks from 1938 and the standard today, the Roman grooved abacus with one and four counters: League of Japan Abacus Associations, “History of the Soroban”, read October 2026; soroban in Japan’s third and fourth grades since the 1989 curriculum: the same, “Soroban in Education and Modern Japanese Society”, read October 2026.
  • The suanpan’s 2 + 5 beads, a rod showing up to 15, 1 jin = 16 liang and base 16, the extra beads in old methods, 三下五除二 (+3 = +5 − 2) as a saying for quick and decisive: Wikipedia, “Suanpan”, read October 2026.
  • The Russian schoty (ten beads a wire, the 5th and 6th of another color), Perelman’s story of the calculator salesmen (as Wikipedia reports it, citing his Arithmetic for Entertainment, p. 51; not found in a text of his we could open), the Cranmer abacus’s soft backing and its use in early grades: Wikipedia, “Abacus”, read October 2026.
  • The 1946 contest (sponsored by Stars and Stripes, Tokyo, Pvt. Thomas Nathan Wood of the 20th Finance Disbursing Section, Kiyoshi Matsuzaki of the Savings Bureau of the Ministry of Postal Administration, 50 numbers of 3 to 6 digits, 1 min 14.9 s against 2 min 0.2 s in the first heat, 4 points to 1): Takashi Kojima, The Japanese Abacus, Its Use and Theory (1954), excerpt on Luis Fernandes’s abacus pages, read October 2026. Kojima dates it 12 November 1946, the Nippon Times (as he quotes it) “Monday afternoon”, and Wikipedia 11 November, so we say only November 1946.
  • Bi-quinary coded decimal, the IBM 650’s seven bits (two “bi” and five “quinary”, exactly one of each set, as an error check): Wikipedia, “Bi-quinary coded decimal”, read October 2026.
  • The Vadodara trial (204 children aged 5 to 7, randomized, three hours a week for three years, gains in arithmetic, none in Raven’s, mental rotation or spatial working memory, gains only for children who began with stronger spatial working memory): D. Barner, G. Alvarez, J. Sullivan, N. Brooks, M. Srinivasan and M. C. Frank, “Learning mathematics in a visuospatial format: a randomized, controlled trial of mental abacus instruction”, Child Development 87(4), 1146–1158 (2016), author’s copy, read October 2026.
  • The imagined abacus held column by column, its computations little affected by verbal interference: M. C. Frank and D. Barner, “Representing exact number visually using mental abacus”, J. Exp. Psych.: General 141(1), 134–149 (2012), abstract, read October 2026 (the paper itself not read).
  • Mental abacus users limited to 3 or 4 columns at a time, beads in sets of 4 or 5 matching the limits of visual attention, and the one-year US trial with no clear advantage: D. Barner and others, “A one-year classroom-randomized trial of mental abacus instruction for first- and second-grade students”, Journal of Numerical Cognition 3(3), 540–558 (2017), read October 2026.
  • The five-year study in China (144 children randomly assigned at the start of first grade, 2 hours a week for 5 school years, better arithmetic and visuospatial working memory, no difference on Raven’s, transfer “quite limited considering the length and intensity of the training”): C. Wang and others, “Training on abacus-based mental calculation enhances visuospatial working memory in children”, J. Neurosci. 39(33), 6439–6448 (2019), read October 2026.
  • About 500 students at the 13th national abacus and mental arithmetic competition, Kathmandu, May 2025: Radio Nepal, “UCMAS Nepal concludes 13th National Abacus and Mental Math Competition”, 25 May 2025, read October 2026.
  • Every number in this topic (the bead patterns, the rules checked for every sum up to 999 + 999, the 90 cases sorted by rule, the fewest moves, the bases, the bead counts, the suanpan’s 15) computed in checks/textbooks/math-further-6-8/39/tmp/abacus_numbers.py.