Twelve problems, one clock, one score. A benchmark you can repeat weekly to see whether practice is actually moving anything.
4 rows × 2 digits, add & subtract
About 5 min
Free, no account
Sample problems
Real problems from this drill, with answers. Nothing to install, no account, no email — work them on paper, on a physical abacus, or in your head.
1
68
+ 93
+ 14
+ 68
?
2
62
+ 72
− 56
− 18
?
3
51
+ 67
+ 37
+ 49
?
4
82
+ 55
− 52
− 10
?
5
48
+ 11
− 40
+ 77
?
6
32
+ 20
+ 54
+ 75
?
7
73
− 44
− 22
+ 26
?
8
25
− 13
+ 72
− 23
?
9
63
− 62
+ 54
+ 58
?
10
58
− 54
+ 52
+ 27
?
11
94
+ 84
+ 90
+ 25
?
12
89
+ 60
− 47
+ 65
?
Full answer key
68 + 93 + 14 + 68 = 243
62 + 72 − 56 − 18 = 60
51 + 67 + 37 + 49 = 204
82 + 55 − 52 − 10 = 75
48 + 11 − 40 + 77 = 96
32 + 20 + 54 + 75 = 181
73 − 44 − 22 + 26 = 33
25 − 13 + 72 − 23 = 61
63 − 62 + 54 + 58 = 113
58 − 54 + 52 + 27 = 83
94 + 84 + 90 + 25 = 293
89 + 60 − 47 + 65 = 167
Practise it now
The same drill, live in your browser, with a virtual soroban and instant marking.
⏱️
Loads a full interactive drill right here on the page. It never asks who you are.
No signup. Nothing saved unless you ask.
About this drill
Practice without measurement drifts. A learner can put in twenty minutes a day for months and have no idea whether they are getting faster, because subjective effort feels identical at every level. A fixed test — same length, same difficulty, same format — turns that into a number.
Treat the score as a trend line, not a grade. Single results bounce around by ten or fifteen percent for reasons that have nothing to do with skill: sleep, time of day, whether the previous problem went badly. Four weeks of readings tell you something; one does not.
How to practise it
1Run it at the same time of day each week, on the same device.
2Do not warm up on the test itself — warm up on a different drill.
3Record both the score and the time; either alone is misleading.
4Change one variable at a time if you want to know what helped.
Questions
What is a good score?
There is no universal benchmark worth quoting — problem width, row count and answer format all change the number. Compare yourself to your own previous result, not to a table.
Should I prioritise speed or accuracy?
Accuracy, always. Speed built on a 70% accuracy base collapses the moment problems get harder; speed built on 100% keeps scaling.
How often should we test?
Weekly at most. Testing more often measures noise, and it eats practice time that would actually move the number.
Does this work as a placement test?
Not really — it measures speed at one fixed difficulty. The placement exam adapts its difficulty and is the right tool for finding a starting level.