Seed Storage

Close-up image of a lettuce seed.

Over time, you may accumulate a great variety of seeds.  It is important to store your seeds properly, to ensure good long-term viability.  Seed storage is pretty simple: keep them in an airtight container, throw in a few silica desiccant packs or some uncooked rice to absorb any moisture, and keep the container in a cool place, such as the back of your refrigerator.  Stored in this way, your seeds should last for 3-5 years with good viability.  Of course, every time you open a seed packet, you are exposing your seeds to atmospheric air and moisture levels, so the fewer times you access the seeds, the longer they will last.  Consider splitting your larger seed packets into smaller-sized portions to prevent over-exposure to the ambient air.  Also consider letting the seeds warm up to room temperature before opening the packets, to prevent moisture from condensing on the cold seeds.

Is it better to store the seeds in the refrigerator or the freezer?  Both places should be fine, but I usually just keep my seeds in the refrigerator, and my seeds remain viable for several years.  For longer term storage (> 5 years), you may want to keep seeds in the freezer, since this is what most professional seed vaults do.

If you want to get extreme about seed storage, then consider the Svalbard Global Seed Vault as the gold standard of seed preservation.  Located on an isolated island near the North Pole, the Svalbard seed vault is buried inside a mountain, where the permafrost should keep the seeds frozen in the event of a power failure.  Inside the vault, seeds are kept at -18 °C in an oxygen-depleted environment.  The seeds are estimated to remain viable for decades or even centuries.

The Svalbard Global Seed Vault, buried inside a mountain near the North Pole.

Calculation of Seed Viability

Even if you store your seeds properly, their viability will still decrease over time.  It can be frustrating to sow a few seeds and find that after a week or 2, they have failed to germinate.  Should you try again and sow more seeds?  Or should you just discard the seeds and buy a fresh pack?  Usually, new seeds can be obtained for only a few dollars, but assuming you want to try and re-sow more seeds, we can apply some basic probability theory to determine your chances of success.

We will assume that the germination conditions are perfect, meaning that the temperature and moisture conditions are ideal.  Let us define success as having at least 1 seed germinate.  Success will depend on 2 factors: 1) the percent viability of the seeds, and 2) the number of seeds sown.  Our equations will also use a third factor, the total number of seeds in the packet.  We do not know the percent viability ahead of time, but we can use the number of seeds sown and the total number of seeds to get a good approximation of the viability.

To do this, let us first ask: “If a packet of seeds has a defined viability, how many seeds do I have to sow to have a 95% probability of success?”  Then we will reverse the question, asking: “If I did not have success after sowing a certain number of seeds, what is the maximum viability of the total number of seeds in the packet?”

The math is a bit complicated, so to get started quickly, just copy and paste the following line directly into the WolframAlpha computational engine, substituting your own values for T and S:

solve for x:  x*ln(x)-(x-s)*ln(x-s) = ln(0.05*t!/(t-s)!)+s and t=T and s=S

x = minimum number of non-viable seeds
t = Total number of seeds in your packet
s = number of seeds Sown

For example, let’s say that I start with a total of 100 seeds, and I have sown 10 of them without any germination success.  I paste this line into WolframAlpha:

solve for x:  x*ln(x)-(x-s)*ln(x-s) = ln(0.05*t!/(t-s)!)+s and t=100 and s=10

and it returns x = 75.8.  This tells me that there is a minimum of 75 non-viable seeds out of the 100 total seeds.  Conversely, there is a maximum of 25 viable seeds out of 100 total seeds, for a viability of 25/100 = 25%.  In other words, starting with a packet of 100 seeds that has 25% viability, there is a 95% chance of getting a successful germination if I sow 10 seeds.  The fact that I was not successful tells me that the viability is likely to be lower than 25%.  This is too low for me, I would likely discard these seeds and buy fresh ones.

Is it possible that in this example, there were actually 90 viable seeds out of the total 100, and I was just unlucky enough to pick out the only 10 non-viable seeds?  Yes, it is possible, but the probability of this is very low (almost 0%).  In fact, the odds of this are 1 in 17.3 trillion!  Anything is possible, but that does not mean it is likely.

Delving Deeper Into Probability Theory

Let’s start with an easy example:  If we have a packet of 10 seeds with 70% viability (7 viable seeds, 3 non-viable seeds), and we sow 1 seed, what is the probability of success (1 germination)?  It is simple: 7/10, or 70%.

Now let’s make it a bit more complicated.  Let’s start with the same 10 seeds at 70% viability, and figure out the probability of success if we sow 2 seeds (success = at least 1 germination).  To get this answer, let’s first reverse the question and figure out the probability of failure (0 germinations).  The probability of failure is 3/10 when we pick the first seed, and then assuming the first seed is non-viable, the probability is 2/9 when we pick the second seed.  Multiplying these 2 probabilities together gives us the total probability of failure: 3/10 * 2/9 = 0.066 = 6.6%.  Then, it is logical that the probability of success = 1 – probability of failure = 1 – 6.6% = 93.3%.  In other words, by sowing 2 seeds instead of just 1, we have increased our probability of success to 93.3% up from 70%.  This is why I always recommend sowing at least a few seeds, to make sure you get at least 1 successful germination.

Following the above logic, let us now consider a slightly more complicated case in which you have a packet of 10 seeds, but this time the viability is unknown, and you have sown 3 seeds and were not successful in getting 1 germination.  Since we don’t know the viability, let us assign the variable ‘x’ = the number of non-viable (dead) seeds in the packet of 10 seeds.  The probability of failure is:

x/10 * (x-1)/9 * (x-2)/8

The probability of success (at least 1 germination) is:

1 – (x/10 * (x-1)/9 * (x-2)/8)

This can be re-written by grouping the numerators and denominators:

1 – (x * (x-1) * (x-2)) / (10 * 9 * 8)

We can generalize this equation by making use of factorials.  Factorials multiply integers in this fashion:  ‘5 factorial’ = 5! = (5 * 4 * 3* 2 * 1).

1 – (x! / (x-3)!) / (10! / (10-3)!)

Even more generally, let’s assign the variable ‘t’ to the total number of seeds in the packet, and the variable ‘s’ to the number of seeds that we have sown:

1 – (x! / (x-s)!) / (t! / (t-s)!)

There!  We have now defined a general equation to figure out the probability of success.  Considering the current example (10 seeds total, 3 seeds sown unsuccessfully), let us now set our requirement for the probability of success to be 95%:

1 – (x! / (x-3)!) / (10! / (10-3)!) = 0.95

Solving for x (by plugging this formula into WolframAlpha) returns x = 4.40285.  Rounding down to 4, this means that there are a minimum of 4 non-viable seeds in the original pack of 10 seeds, so maximum viability is 6/10 = 60%.  In other words, starting with a packet of 10 seeds at 60% viability and sowing 3 of them would be expected to result in success 95% of the time.  The fact that we were not successful indicates that the actual viability is less than 60%.  Our calculation gives us the maximum level of viability; it does not give us the actual viability, since this could be much lower than 60% (it could even be 0 % if all the seeds are non-viable).  A maximum viability of 60% is not great, but if these were my seeds, I would probably sow a few more to see if I could get one to germinate.

Now that we know that the seeds have a maximum viability of 60% (4 non-viable seeds out of 10), let’s double check our work to ensure that it makes sense:

probability of success = 1 – (4! / (4-3)!) / (10! / (10-3)!) = 29/30 = 96.6%…Close to 95%, so it checks out.

Let’s do one final example, using the scenario from above involving a packet of 100 seeds of unknown viability, and 10 seeds sown without a success:

probability of success = 1 – (x! / (x-10)!) / (100! / (100-10)!) = 0.95

Solving for x (by plugging this formula into WolframAlpha) returns x = 75.3046.  Rounding it down to 75 non-viable seeds results in a maximum viability of 25/100 = 25%, the same result we got above.

A final note on plugging equations into WolframAlpha.  Some equations, especially those involving factorials, are not easily solved by the program.  In this case, it is best to use Stirling’s approximation of factorials which uses natural logarithms (ln):

ln(n!) ≈ n * ln(n) – n

The Bottom Line

Seeds do not always germinate with 100% efficiency.  If you want to get a successful germination, sow at least 2 or 3 seeds to ensure that you aren’t using a non-viable seed, and if they all sprout, just thin down to one plant if desired.  If 2 or 3 seeds fail to germinate (and you are confident that the germination conditions are ideal), you can estimate the maximum viability of your seeds using the tips above, or just get some new seeds.