Crop Growth: Points and Random Ticks

Verified: Java 26.3 · 2026-09-19 · Java modelled; Bedrock's random tick speed differs and is declared, not computed

A wheat farm takes about 24 minutes, and almost every part of that sentence is doing work. This guide derives the number the crop growth calculator reports: where growth points come from, why one plant's neighbours matter, what the floor function quietly forgives, and why the average is the least useful of the three times the tool prints.

Nothing grows on a timer

Every block in a loaded chunk sits in a 16×16×16 subchunk of 4,096 blocks. On each game tick — there are 20 a second — the game picks randomTickSpeed blocks at random from that subchunk and gives each a random tick, and the same block can be picked twice. With the Java default of 3, a given block is picked 3 times in 4,096 on average, which works out to one random tick every 68.27 seconds.

A crop does nothing between random ticks. When one lands, it rolls: succeed and it advances a stage, fail and it waits for the next tick. Wheat, carrots, potatoes and melon and pumpkin stems have 8 stages, so they need 7 advances; beetroot has 4 stages and needs 3.

Growth points

The roll is not fixed. It comes from a score the game builds out of the 3×3 of blocks under and around the plant:

points = 1 + (moist ? 3 : 1) + 0.75 per moist neighbour + 0.25 per dry neighbour

The tile directly under the crop counts at full value, each of the eight around it at a quarter, and anything that is not farmland contributes nothing. A perfectly watered, uncrowded plant scores 1 + 3 + 8 × 0.75 = 10, the maximum. The same plant on dry farmland scores 1 + 1 + 8 × 0.25 = 4.

Then the crowding rule takes its cut. If the same crop sits on any diagonal, or on both the north-south and the east-west axis, the score is halved. That single line is the whole rows-versus-solid argument: a solid field and a checkerboard trip it, alternating rows and a single row do not.

The score becomes a probability through

chance = 1 / (floor(25 / points) + 1)

so 10 points give 1 in 3, and 5 points give 1 in 6. One advance then takes 4,096 ÷ (3 × chance) game ticks on average — 204.8 seconds at 1 in 3 — and a crop needs one of those per advance.

Four layouts, one plot

Wheat on moist farmland, 9×9, at the default tick speed. The last column is the plot's output, not one plant's, so it counts the plants each layout fits.

Rows win twice: fastest per plant, and most out of the plot. The solid field stays close on throughput because 81 plants at half speed is nearly 45 at full speed, which is worth knowing if floor space is what you are short of. The checkerboard gets the worst of both.

What the floor forgives

Because the formula floors 25 ÷ points before adding one, whole ranges of scores collapse onto the same roll. Anything from 8.5 to 10 points rolls 1 in 3; anything from 6.5 to 8.25 rolls 1 in 4. In practice that means a plant can lose one or two of its eight farmland neighbours for free — 10, 9.25 and 8.50 points are the same 1 in 3 — so the water hole in the middle of the plot, a torch base, or a path tile beside the row costs nothing at all to the crops around it. The third missing neighbour is where it lands: 7.75 points, 1 in 4, and wheat goes from 23 min 54 s to 31 min 51 s.

The same floor sets the real worst case. The lowest score the arithmetic allows is 1.0, but you cannot build it: a crowding neighbour has to be standing on farmland itself, so it pays 0.25 back before the halving. The floor in the game is 1.125 points, a roll of 1 in 23, which is two crops diagonally adjacent on dry farmland with nothing else around them.

Why the average is the wrong number for a clock

Moist rows of wheat average 23 min 54 s, but a field is not an average. The time to maturity is the wait for 7 successes out of a stream of rolls, which is a long-tailed thing: half the plants are done by 22 min 46 s, and one in five is still growing at 30 min 59 s. An automatic farm on a timer wants that last figure, not the mean, because it harvests everything at once and every unripe plant is a wasted cycle.

Two of this site's numbers differ from the wiki's on exactly this ground, and both are published with the disagreement attached rather than quietly reconciled. The first is the per-plant rate. The tutorial prices a wheat crop at 6.591 × growth chance harvests an hour; that coefficient is the arithmetic for eight advances, and wheat makes seven — as the same tutorial states two sections earlier, where it counts wheat's eight stages. The right coefficient is 7.533, which is what this site uses, and at 1 in 3 the difference is 2.51 harvests an hour against 2.20.

The second is the timer itself. The tutorial gives the four-fifths-matured point as "precisely 31 minutes and 3.14 seconds". Computing it here two ways — an exact discrete calculation over game ticks, and a continuous one — gives 1,858.75 and 1,858.64 seconds, about 4.4 seconds earlier and within a twentieth of a second of each other, so the gap is not a discrete-versus-continuous artefact. The cause is not established. This site prints 30 min 59 s, says it is its own arithmetic, and leaves the wiki's figure standing beside it.

What this does not model

Bone meal is not in any of it: it forces advances directly, ignoring points and random ticks, and a stack of it makes every number here irrelevant. Light is a gate rather than a dial — a crop needs light level 9 at the plant block to grow at all, and more light does not make it faster. Bees pollinating a crop advance it a stage for free, which the tool does not count. Melon and pumpkin stems are priced to maturity only; the mature stem then spawns fruit on a separate, slower check that only one of this page's two sources publishes numbers for, so this site does not price it. And the two ways of counting a random tick — the sources' 4,096 ÷ (3 × chance) and an exact per-game-tick calculation that allows for the same block being picked twice — differ by eight thousandths of a percent, which never reaches the second the page prints.

Sources

The points rule, the roll, the speed-level range, the per-random-tick table, the stage counts and both of the figures held above are on the Minecraft Wiki's crop farming tutorial; hydration is from its farmland page and the 4,096-block subchunk with its default random tick speed from its tick page, all read 2026-09-19. Every constant was checked the same day against the crop growth calculator at minecraftmaps.com/tools/crop-growth-calculator, which publishes the same 3-and-1 tile values, the same quarter weight on neighbours, the same halving rule and the same times. The arithmetic on top — the distribution, the plot throughput and the two corrections — is this site's, and every number printed on this page and on the tool is asserted in tests/mirror_crop-growth-calculator.py, which rebuilds the wiki's own 18-cell table from the formula rather than copying it.