Experimenting is legitimate; overstating the result is not. To learn something meaningful from your trial, a few simple rules are enough. They do not substitute for a laboratory study, but they move you from “I tried it, I think it worked” to “I measured this, and this came out”. That difference is large.

1. Keep a control group

Split plants of the same species and similar condition into two groups. Apply to one, not to the other. Without a control you cannot know the cause of any change you see — season, watering, temperature, light, soil and the plant’s own growth stage all act at once.

The two groups should be as similar as possible: same species, similar size, same pot type, same soil, same location. Compare two plants at different windows and what you measure may be the difference in light.

2. Assign plants to groups at random

Do not decide which plant goes where by looking at it. The moment you think “this one is weaker, put it in the treatment group”, the trial is broken. Flip a coin, or number them and draw. This is not a formality: a choice you make without noticing determines the result in advance.

3. Change one variable

Run a frequency trial and change fertiliser in the same week, and you will not know which to credit. One thing at a time. This is the most frequently broken rule and the one that ruins the most results.

4. Decide what you will measure before you start

“Did it get better?” is a vague question and everyone finds the answer they wanted. “Number of yellowing leaves”, “number of new shoots”, “lesion diameter (mm)”, “plant height (cm)”, “harvest weight (g)” are measurable things.

Choose and write down the measure before you begin. Choosing afterwards pushes you toward self-deception: look at ten things and one will look good by chance, and that is the story you start telling.

5. Fix the application conditions and write them down

In a sound trial the following directly affect the outcome and all should be recorded:

Hz value and waveform
Distance between device and plant (centimetres)
Level (device percentage, or a decibel reading if you have one)
Minutes per day, and at what time of day
How many days it ran
Environment: room / greenhouse / field, and background noise

Distance matters particularly: sound pressure falls quickly with distance, and doubling it weakens the signal markedly. “I put it nearby” is not a measurement.

6. Keep dated, consistent records

A simple table suffices: date, applied or not, measured value, notes. Take photographs — same angle, same distance, same light, ideally the same hour. Memory is unreliable; notes are reliable. Recalling two weeks later that “I think it was greener” amounts to not having measured at all.

7. Long enough, and enough of them

A three-day trial with three plants tells you nothing. Plants change slowly; a measurable difference usually takes weeks. The more individuals per group, the less chance can distort the result. If you can, aim for at least five plants per group, ten if possible.

8. Watch for expectation bias

When you want something to work, your tendency to see it working increases. This is not a character flaw but a cognitive tendency everyone has, and professional researchers take specific measures against it.

The simplest measure available to you: have someone other than the person running the trial take the measurements, and keep them unaware of which plant is in which group. Just number the pots; keep the group assignment on a separate sheet. This is called blinding, and on its own it multiplies the value of your result.

9. Remember regression to the mean

A plant in very poor condition tends to look better at the next measurement even if you do nothing; the worst moment is usually the worst moment, and what follows is better. Starting an intervention at the worst moment and crediting the subsequent improvement to it is the most common error in this field. A control group exists precisely to catch this.

10. Write down the negative result too

A trial can end with “nothing came of it”, and that is not a worthless outcome — on the contrary, it is the most frequently hidden and therefore the most valuable one. Be honest in your own record, even if you never show it to anyone.

A record template

Open these columns in a notebook or spreadsheet and fill them daily: Date · Group (treatment/control) · Plant no · Hz · Distance (cm) · Level · Duration (min) · Measurement · Note · Photo taken. Ten days later you have data worth discussing; and when you do not, you at least know what you failed to do.