A student-led project compared four-year-old sweet corn seed, seed packed one year ago and seed packed for 2015. For each kind of seed, 50 seeds were planted in two trays. Each tray was a replicate.
It sounds like an experiment so far, right?
The most obvious result was that while some of the four-year old seed grew, the germination was only 14% in each tray. Germination of the other two sweet corns in the test were 70% or higher.
Related: Solar Panels, Field Data Inspire Farmer's Corn Yield Trials
Even with sloppy methods, it would be obvious that four-year-old sweet corn doesn't hold and emerge as well as one-year-old or newly packed sweet corn seed.
The rub comes in on the other two. The one-year-old seed actually emerged 7% better than the new seed. The students were all ready to declare that it was superior to the new seed.
Hold the phone, students. There were a lot of variables at work here. It was a classic case of many causes of experimental error. While statistical analysis wasn't attempted, it's most likely those two would not come out statistically different at any meaningful level. What is likely obvious is that both are better than planting seed older than one year – especially four-year-old seed.
Related: For Planting Corn, Soil Temperature Of 50 Degrees Is Magic Number
What kind of "experimental error" – or background noise as Bob Nielsen, Purdue University Extension corn specialist, calls it – could be at work here?
Here are just a few possible sources of error the students identified.
1. Two different people planted the corn: One planted the oldest corn, and another person planted the newer seed.
2. Containers. Containers in which the seed was planted weren't identical. Two were deeper than the others, likely resulting in deeper planting.
3. Brand difference. All three brands of seed were different.
4. Location. The location in the greenhouse, mainly the micro-environment, was different for each. It could have affected how much sun and heat one got versus another.
5. Watering was inconsistent. The trays weren't watered uniformly every day and some dried out more than others before being watered before corn emerged.
6. Different groups counted the plants. It's best to have the same person making counts to be consistent.
Any of these six factors could have introduced enough error to affect the results, especially among the newer seed. What did the students conclude after discussing possible errors? Don't plant very old seed, but planting one-year-old seed probably will deliver about the same results as planting new seed.