To test seed germination, count out 100 seeds at random, germinate them between damp cloth or in a tray of moist sand, then count only the normal seedlings on the day that crop is due. The count is your germination percentage. It costs a handful of seed, a cloth and about a week, and it tells you whether an acre of fertilizer, labour and weeding is about to be spent on seed that cannot repay it.
MAAIF's extension manuals put the test a week before planting, which is the right timing because it leaves room to buy different seed. What the resulting figure means, and why a laboratory number is always higher than your field, is on seed germination rates. This page is the procedure, and it sits with our other farm input guides.
When to Run a Germination Test on Farm Seed
Not every pack needs testing. Six situations make it worth the week.
Seed you saved yourself. Your own store has no test date and no label, so a count is the only figure you will ever have for it. Seed over a year old, including a pack you bought last season and did not finish. Seed with no germination percentage printed on it, which means either it never went through certification or the packaging is faked, and in both cases you are planting blind. Seed bought at an unusual discount or from a source you have not used before. Seed you are about to commit a whole acre to, where the arithmetic of a total failure is worth a week of your time. And seed from the same lot as a crop that came up patchy, because that is the only way to settle whether the seed or the seedbed was at fault.
There is one honest exception. Where you have only a small quantity of seed and no realistic way to replace it, a tropical extension service advising commercial growers says outright that you may as well skip the test and plant what you have. Using ten of your last fifty seeds to learn that the remaining forty are poor helps nobody.
What a Germination Test Needs on a Smallholder Farm
Nothing in this test needs buying, which is the point of it.
Choose the method by seed size. Cloth suits small seed, tomato, cabbage, onion, pepper and the rest of the vegetables, because you can see every seed and spot the first white root as it breaks the coat. Sand suits large field crop seed, maize, beans, groundnuts and cowpeas, because sand gives good contact, the shoot pushes up into the light, and you end up with a seedling substantial enough to judge properly. FAO notes that sand is the preferred medium exactly when fungal growth makes a seedling hard to evaluate on paper.
How Many Seeds to Test, and How to Pick Them
A laboratory uses 400 seeds, split into four replicates of 100. On a farm, 100 seeds is the number to aim for, for the simple reason that the arithmetic disappears: each seed is one percentage point, so the count is the answer. Where a vegetable sachet is too small to spare 100, use 50, or 20, or 10 if that is all the sachet allows.
Understand what the smaller sample costs you. The reliability of the test is relative to the size of the sample, so a 10 seed test moves in steps of 10 percentage points and cannot tell 85 from 90. Where the decision is expensive and the sample has to be small, run two or three separate tests rather than one, and take the picture across them.
How you choose the seeds matters more than how many, and this is where most farm tests go wrong. Take them at random from well mixed seed and do not cull anything. An extension service publishing the wet towel method is explicit: do not remove damaged, discoloured or light seeds, because doing so biases the test. Picking out the best looking seed is the natural thing to do and it produces a figure that describes the seed you chose rather than the seed in the bag.
Two more rules. Test the pack you are actually going to sow, not a different pack of the same variety from the same shop, because lots differ. And where you suspect a seed borne fungus, split the sample into groups of 25 or 50 in separate containers, so one rotting seed cannot take its neighbours down with it and distort the count.
The Rag Doll Test: Germinating Seed Between Damp Cloth
The rag doll is the paper towel test under its farm name, and the whole method is nine steps.
Two things go wrong, and each has one cause. Seeds that rot mean the sample was too wet from the start. A sample that dried out means the container was not properly sealed. Neither gives you a germination figure, so start again rather than counting what is left.
The Sand Tray Test MAAIF Recommends for Field Crops
For maize, beans, groundnuts and the other large seeded field crops, MAAIF's extension manuals set out a sand method instead, and it is five steps.
Sand has a real advantage over cloth for field crops beyond just holding a bigger seed. The seedling grows up out of the medium into the light, which develops the leaves and cotyledons properly, and it is those structures you are judging. A maize seedling curled inside a damp towel is harder to call than the same seedling standing in sand.
How Long to Leave the Seed Before You Count
The waiting period is set by the crop, not by your patience. A laboratory makes two counts: a first count after a few days, which says how fast the lot is moving, and a final count on a fixed day, which is the figure that gets reported. The international rules fix both for each species.
| Crop | First count | Final count |
|---|---|---|
| Pearl millet | Day 3 | Day 7 |
| Maize | Day 4 | Day 7 |
| Barley | Day 4 | Day 7 |
| Wheat | Day 4 | Day 8 |
| Cowpea | Day 5 | Day 8 |
| Sorghum | Day 4 | Day 10 |
| Groundnut | Day 5 | Day 10 |
| Oats | Day 5 | Day 10 |
| Rice | Day 5 | Day 14 |
MAAIF's own field instruction for beans is a single count at day 7, which sits sensibly between the cowpea and groundnut figures above. Groundnut is tested with the shells removed.
Do not extend the test to flatter the result. Seed that germinates a week after the final count day is scored as ungerminated in a laboratory, and it is worth no more than that in your field, because your seedbed will not wait for it either. Slow emergence is itself a signal: a lot that limps to 88% over a fortnight is a weaker buy than one that reaches 88% on day seven.
Sorting the Result: Normal Seedlings, Abnormal Seedlings and Dead Seed
This is the step that decides whether your number is worth anything, and it is where a farm test most often overstates the seed. Only normal seedlings count. A seed that sprouted and then produced something deformed is a failure, not a pass.
Write the three counts down as you go, one line per container, with the batch and the total number of seeds set. MAAIF's record sheet is exactly that: batch, total seeds, normal seedlings, abnormal seedlings, and rotten or diseased or unviable seeds. If you are testing two packs, the written record is the only thing that will keep them straight.
Working Out the Percentage and What It Tells You to Plant
The calculation is one division. Divide the number of normal seedlings by the number of seeds you set, then multiply by 100, and round to a whole number. With 100 seeds the count is the percentage.
Then act on it. MAAIF's thresholds are clear and they are about seed rate rather than about panic.
| Normal seedlings from 100 | What to do |
|---|---|
| 90 or more | Plant at the recommended rate |
| 85 to 89 | Raise the rate to compensate |
| Below 85 | Reject the seed |
The adjustment in the middle row is MAAIF's own arithmetic. Divide 100 by your germination percentage, then multiply by the recommended seed rate for that crop. Worked through on beans, where 30 kilograms per acre is the recommended rate and the count came out at 85%: 100 divided by 85, multiplied by 30, gives 35.3 kilograms per acre. The extra 5.3 kilograms buys back the plants the weak seed will not deliver. Use your own crop's recommended rate rather than the bean figure, and the full arithmetic for other spacings is on our seed rate calculator.
Take the bottom row seriously. Below 85%, MAAIF's instruction is rejection rather than a bigger adjustment, and there is a reason beyond caution: sowing a very weak lot heavily gives you a thin stand that is also crowded in places, which is worse agronomy than a thin even one. The one place a lower threshold is defensible is a transplanted vegetable crop where the seed is already bought and cannot be returned, and some extension services treat 70% as the point to buy new seed for those. Why the thresholds differ is set out on germination rates.
What to Do About a Poor Germination Result on Bought Seed
A bad count on seed you paid for is a claim, and claims have a clock on them.
MAAIF's instruction is to report back to the stockist, the extension worker or the seed company within 10 days after planting where germination has failed. If you tested before planting, which is the point of this page, report as soon as you have the count rather than waiting to plant and fail. Go to all three rather than one: a word to the dealer alone leaves no record anywhere it can be acted on.
Keep the evidence, and keep it in one place. The packet, the label, the lot number, the receipt, your written tally, and the sorted seedlings themselves laid out on their paper. Photographs cost nothing and a photograph of three labelled piles beside a pack with its lot number visible is a far stronger position than a description.
Then do three things before you buy again. Run a second test on a fresh sample from the same pack, because one test on a small sample can mislead and two agreeing tests are hard to argue with. Do not buy another pack carrying the same lot number from the same shop, since the lot is the batch and the batch is what failed. And check the class, the seal and the germination figure on whatever you buy instead, using the checklist on how to choose quality seed.
When the Seed Passes on Cloth and Fails in the Field
The test is more useful as a diagnostic than as a pass mark, because it cuts the possible causes of a failed stand into three groups and tells you which one you are in.
It fails on cloth and failed in the field. The seed. You have a number, a lot and a claim, and the section above is what to do with them.
It passes on cloth and failed in the field. Not the seed. Something in the seedbed or the nursery stopped seed that was perfectly capable of growing. The usual suspects, roughly in order of how often they turn out to be the answer: planting depth, where too deep delays or prevents emergence and too shallow leaves the seed without the moisture it needs to start; soil moisture arriving and leaving between showers; a crust forming after heavy rain and trapping the shoot; poor contact between seed and soil in a cloddy tilth; damping off in a wet nursery; and ants, birds and rodents. A tropical extension service puts this plainly: seed that germinates in the test but fails to emerge in a seedling tray points to a disease or to conditions inhibiting germination rather than to the seed.
It passes on cloth, comes up in the field, and comes up weak and uneven. Vigour, and this is the honest limit of the method. No cloth or sand test on a farm measures vigour, and a laboratory needs separate tests for it. Two lots at 90% can produce very different stands in a dry seedbed. What you can read as a proxy is speed: a lot that hits its count on the final day rather than early, or that germinates raggedly across several days, is telling you something a single percentage cannot.
One rule keeps all three readings honest. Test the pack you actually sowed, from seed left in that same pack, not a second pack of the same variety. Otherwise you have tested something else and learned nothing about your field.
Testing Seed Germination: Questions Farmers Ask
How many seeds do I really need for a reliable test?
One hundred, if the pack allows it, because each seed is then one percentage point and the count is the answer. Fifty works. Twenty and ten work but move in steps of five and ten percentage points, which is too coarse to tell 84 from 90, and 85 is where MAAIF's decision sits. Where you are stuck with a small sample, run two or three tests and read them together rather than betting an acre on one.
Can I use cotton wool, a cloth or a plate instead of paper towel?
A clean cotton cloth works and is what gives the rag doll its name. The requirements are that the medium holds moisture evenly, lets air in, and lets you lift the seedlings off without tearing the roots. Cotton wool fails on the last of those, because fine roots grow into it and break when you pull them out. An open plate fails because it dries out within a day. Fine sifted sand in a container with drainage holes is the alternative MAAIF uses and the one to prefer for large seed.
Should I plant the seedlings from my test?
For field crops, no. They have been handled, their roots have been disturbed, and a hundred of them will not fill any part of an acre worth planning around. For a few expensive vegetable seeds where the test doubles as pre sprouting, the strongest seedlings can be pricked out into a nursery tray with care, and the ones you would classify as abnormal should be discarded rather than nursed. Nursery method is on tomato nursery management.
What does testing cost, and can a laboratory do it for me?
On the farm it costs the seeds you use plus a cloth, which is why it is worth doing on any pack you are unsure about. Formal laboratories exist here, some of them holding ISTA accreditation, and a laboratory test is a step every certified batch already went through before it reached the shelf. What those laboratories are not set up to do is turn round one farmer's sachet on demand, and any fee for taking a private sample varies and has to be agreed with them directly. So for a decision you have to make in a week, the cloth in your own kitchen is the instrument that is actually available.
Will a germination test tell me whether my seed is fake?
Partly, and it is the strongest cheap signal you have. Adulterated or repackaged seed usually germinates badly, so a poor count on a sealed, certified looking pack is real evidence. What the test cannot detect is the other kind of counterfeit: seed of a different variety, or grain of the right crop from an unknown source, which can germinate beautifully and then grow into something that is not what you paid for. That failure only shows in the field, which is why the seed class and the lot number matter alongside the count.
My test came out higher than the figure on the packet. Is that possible?
It happens, and the usual explanation is counting rather than seed. Check whether you counted every sprouted seed instead of only the normal seedlings, because that is the mistake that inflates a farm count by several points. Sample size is the other candidate: a lucky 20 seed sample can easily read 95 on a lot that is really 85. A genuinely higher figure is possible where the printed number is a conservative minimum for the class rather than the measured result for that batch.
Do I have to test every pack, every season?
No. Fresh certified seed from stock that turns over, bought sealed with a lot number and a germination figure printed on it, is the case where you can reasonably plant on the label. Test the saved seed, the old seed, the bargain, the unlabelled, and anything going into a whole acre at once. That is a short list and it covers almost every failure worth avoiding.
None of this replaces buying seed fresh in the first place, and freshness comes down to which shelf the pack sat on and for how long. Our input supplier directory is where to start looking for dealers whose stock moves, and where maize seed is sold in Uganda sets out the channels for the crop where a failed stand hurts most. One more thing worth doing before you need it: tell a local extension officer that you run these counts. An officer who has seen your method, and who will look at your three piles of seedlings when a batch goes wrong, turns a private tally into something a seed company has to answer.
