Stitch Distribution Calculator

Almost every knitted or crocheted garment changes stitch count somewhere. A hat crown closes from ninety-six stitches down to nothing, a sweater body narrows into the yoke, a sleeve gathers into a cuff, and ribbing opens out into the body above it. In every one of those the pattern says the same thing: decrease evenly across the row. What it does not say is how, and that is the entire problem — bunch the changes together and you get a visible pucker or an unintended dart where you wanted a smooth taper.

Why one interval can't be the answer

The obvious move is to divide. You have 100 stitches, you want 73, so that is 27 decreases, and 100 ÷ 27 is 3.7 — call it "every 3 stitches" and get on with it. Work that row and you finish with 67 stitches, not 73: decreasing every 3 stitches across 100 of them is 33 decreases, six more than you asked for. Round the other way, to "every 4", and you get 25 decreases and 75 stitches. Neither interval is the answer, and no third one is either, because 3.7 is not a number of stitches.

This is not a hypothetical failure. It is what the calculators that exist for this actually do — take the difference, divide, floor it, print "every N stitches" — and the remainder the floor throws away is neither distributed nor mentioned. It only shows up when the row is worked and the count at the end is wrong.

The real answer uses two group sizes, not one. Take the two whole numbers either side of 3.7 — 3 and 4 — and mix them in the one ratio that makes the total come out exactly right: 19 groups of 4 and 8 groups of 3. That is 19 × 4 + 8 × 3 = 100 stitches used and 19 + 8 = 27 decreases worked. It is the same arithmetic a computer uses to draw a straight diagonal line on a square grid of pixels, for the same reason: you have a fractional ideal spacing and only whole units to build it from, so you alternate between the two nearest whole values in the proportion that averages out.

The trap: a group is not its plain stitches

Knowing the group sizes is not the same as knowing the instruction, and the step between them is where this goes wrong quietly. A group of 4 does not mean "work 4, then decrease". A decrease consumes two stitches and gives back one, so a group of 4 is 2 plain stitches and then the decrease — the decrease is eating the other two. Reach for the increase rule by mistake and write "work 3, decrease" instead, and every group swallows one stitch more than it should: 27 groups, 27 stitches over, and the row would need 127 stitches instead of the 100 you have.

An increase is the mirror image and the arithmetic runs the other way. A single increase is worked into one existing stitch and turns it into two, so it consumes one stitch, not two, and the plain count is the group size minus 1. Getting these two the wrong way round produces an instruction that reads perfectly well, is off by exactly one stitch per group, and fails for the first time when a person is halfway along the row with a needle in their hand.

So for 100 down to 73 in the round, the written form is:

[sc 2, sc2tog] × 19, [sc 1, sc2tog] × 8

or in knitting, the identical distribution with different names on it: [k2, k2tog] × 19, [k1, k2tog] × 8.

The two totals, and why they are shown

Every instruction this tool emits has to satisfy two independent sums, and both are printed directly under the answer rather than left implicit:

• Stitches consumed must equal the row you have. 19 × 4 + 8 × 3 = 100.
• Stitches produced must equal the row you want. 19 × 3 + 8 × 2 = 73, since each decrease group gives back one fewer than it took.

The reason to show both is that they fail separately. An instruction with the wrong plain count still produces the right number of decreases — it satisfies the second sum and breaks the first. An instruction with the wrong number of groups can consume the whole row and still land on the wrong count. Only both together pin the answer down, and once you can see them you do not have to trust the tool: you can check it in your head before you pick up the hook.

Flat rows and working in the round

In the round the groups wrap continuously and the instruction covers exactly one round — there is no seam for a decrease to land awkwardly on, so the whole count is available for shaping.

A flat row is different. Working a decrease into the very first or very last stitch of a row makes the edge ragged and the seam hard to sew, so the convention is to keep one or two stitches plain at each end and shape only between them. That is a real input, not a cosmetic one, because it changes the distribution: those edge stitches come out of the count the groups are spread over. The same 100 → 73 decrease worked flat with one edge stitch at each end spreads over 98 stitches instead of 100 and comes out as sc 1, [sc 2, sc2tog] × 17, [sc 1, sc2tog] × 10, sc 1 — a different mix of the same two group sizes. Set the edge count to 0 if you are shaping right to the edge anyway.

Which decrease, and the one increase that breaks the arithmetic

The knitting instructions use k2tog and kfb, and crochet uses sc2tog and a doubled stitch. Substituting a different decrease is free: ssk consumes two stitches and yields one exactly as k2tog does, so swapping it in for a left-leaning pair changes how the fabric looks and nothing about the counts. Use whichever your pattern calls for.

Increases are not so simple, and this is worth knowing before you substitute. kfb works twice into one existing stitch, so it consumes one — which is what the group sizes here assume. m1 (make one) is different: it is worked into the strand between two stitches and consumes none of them. If you swap m1 in, every group keeps all of its stitches and the plain count becomes the full group size rather than one less. The stitch counts still come out right, but the instruction has to be written differently — "k4, m1" where this tool says "k3, kfb". Same spacing, same result, one more plain stitch per group on the page.

Why the groups are written as blocks

Strictly, the most even possible arrangement alternates the two group sizes throughout the row — 4, 4, 3, 4, 4, 3 and so on. Nobody writes a pattern that way, because it is unreadable and impossible to keep your place in. The convention, and what this tool emits, is to write all the groups of one size and then all the groups of the other. Since the two sizes differ by exactly one stitch, the difference between the two arrangements is one stitch of spacing at the transition — invisible in the finished fabric, and worth trading for an instruction you can actually follow while counting.

When it won't fit in one row

There is a hard limit on how much shaping a single row can absorb, and the tool refuses rather than quietly answering a smaller question. Each decrease eats two stitches, so you cannot work more than half as many decreases as you have stitches — 40 stitches allows 20 decreases and no more, taking you to 20. Increases are bounded too, at one per stitch, since each has to be worked into a stitch that exists: 30 stitches can become 60 and not 61. Beyond either bound the shaping has to be spread over two or more rows, which is what real patterns do anyway — a hat crown decreases over eight or nine rounds, not one.

What it assumes

One decrease or increase per group, which is what "evenly" means in a pattern instruction and covers the ordinary cases. Double decreases (three stitches into one) and multiple increases worked into a single stitch are outside it. Every group is a plain run followed by one shaping stitch, which is the standard written form; if your pattern centres the change inside the group instead, the counts are identical and only the reading order differs. And the stitches are counted, not measured — gauge, yarn weight and hook or needle size change how big the finished row is, but not how the changes are spaced across it.