A seed packet says 70 days. Seventy days from what?

The one sum in gardening that has to be right

You are standing in front of a seed rack in May. The packet says 70 days. Your first frost is around the tenth of October. Seventy days is comfortably inside that, so you buy it.

Seventy days from what, though?

There are two possible answers and the packet often does not tell you. It can mean seventy days from the moment you put the seed in the ground. Or it can mean seventy days from the moment you set an already-started plant outside — which quietly skips the several weeks that plant spent growing on a windowsill or in somebody's greenhouse.

Those are not close to the same thing, and the arithmetic you are doing in front of the rack only works if you know which one you are holding.

So we measured our own data, because we store them separately

Our plant records keep two different fields for this on purpose: days from sowing, and days from transplanting. A source that does not say which one it means cannot be filed in either without somebody deciding, and we would rather leave a blank than guess.

Of 156 crops, 104 carry a from-sowing figure, 63 carry a from-transplant figure, and 56 carry both. Those 56 are the interesting ones, because they let us compare the two numbers for the same plant instead of arguing about them in the abstract.

On the 49 where both figures are directly comparable:

• The median gap is 22 days. Three weeks, not the "six weeks" that gets repeated.
• 20 of the 49 differ by 28 days or more. A month, on the number you plan your season with.
• The widest is petunia at 59 days — 80 from seed, 21 from a transplant.
• Then eggplant at 42, celosia and salvia at 40, ground cherry and oregano at 37.

Six weeks turns out to be near the top of the range rather than the middle. We had been saying six weeks ourselves before we counted.

The number nobody can give you honestly, including us

Here is the part that should temper all of the above. Roughly 58 of our records say, in their own notes, that the source never stated which starting line it meant. They were filed under one anchor anyway, because a figure with an unstated basis is still better than no figure — but the record admits it is asserting something the page did not say.

And we cannot count those precisely either. Scanning for the narrow phrasing finds 35 of them; scanning for the broad phrasing finds 64. The honest answer is "roughly 58, and the exact number depends on how you ask" — which is a slightly ridiculous sentence to publish, and truer than picking one.

The distribution is stable though, and it is dominated by one crop: 24 of them are onion. If you have ever found onion day-counts confusing, that is not you.

Two varieties of one crop, on different start lines, side by side

It gets worse inside a single crop. 13 crops are split across their own varieties — some filed from seed, some from transplant, because that is how each individual source published it.

Lettuce is 2 varieties from seed against 52 from transplant. Melon is 1 against 48. Watermelon is 30 against 17. Onion runs the other way, 38 against 1.

So two varieties of the same crop can sit next to each other with numbers counted from different moments, and nothing on the page tells you. That is not a failure of any one source. It is what happens when a whole industry prints one number for two measurements.

Where we are still wrong, and it is on our own pages

We built two fields specifically so this distinction could not be lost. Then we lose it at the last step.

Our export collapses both fields into a single "days to maturity" string. It keeps the basis for individual varieties, but drops it entirely for the crop as a whole — and the crop page then prints a sentence like "It matures in about 52–90 days" with no indication of what it is counted from.

Go and look at our tomato page right now and that is what it says. Every tomato variety we hold is recorded from transplant. A reader who takes 52–90 as sow-to-harvest and counts back from their frost date is out by weeks, on our site, because of us.

That is the whole point of this post arriving as an admission rather than a lecture. We identified the problem, built the structure to fix it, and still ship the flattened version to readers. It is queued.

Two numbers that cannot both be right

Measuring the 49 also turned up two records that are impossible on their face, and we would rather print them than tidy them away.

Bok choy has the same figure for both — 60 days from seed and 60 from transplant. Those cannot both be true; transplanting skips real weeks.

Watermelon has the transplant figure longer than the from-seed one — 87 against 80 — which is backwards by definition. Watermelon is also one of the crops whose sources never stated a basis, so this is very likely two figures on different start lines being compared as if they were not.

Both are now flagged for re-sourcing. Neither has been deleted, because an odd-looking number that came from a real page is evidence of something, and deleting it destroys the evidence along with the oddity.

What to actually do in front of the seed rack

Assume the longer one unless the packet says otherwise. If a count is from transplant and you treat it as from-seed, you will start too late and run into frost. The other mistake costs you nothing but an early harvest.

If the packet says "days to maturity" with no anchor and the crop is normally started indoors — tomato, pepper, eggplant, melon — it is almost certainly from transplant. Those are the crops where the two numbers are furthest apart, which is exactly why it matters most there.

And add a margin. Every figure of this kind is from a trial in somebody else's conditions. Our own frost dates come with a ten-to-ninety percent band for the same reason: a single number for a biological process is a convenience, not a promise.