Let me admit at the outset that my sample is small. However, it's perhaps realistic, given that the home gardeners I'm trying to persuade to do research don't live on research farms. Small scale experiments can generate useful data, and enough such experiments might help us reach tentative conclusions about what we're testing.
…
What makes an experiment scientific? You'll need a control group, you'll need to eliminate as many variables as possible, you'll need to report what actually happens, not what you want or expect the results to be, and you or others will have to repeat the experiment.
The control group in my experiment is the four Early Girl plants that don't get tea. I know from experience that Early Girl is susceptible to late blight. If the disease shows up in my garden this year, chances are good those four plants will catch it. If they do and the four tea-treated plants don't, the case for tea as a control is strengthened. If none of the eight plants in the test catches the disease though, I can't make a claim for the tea.
By way of eliminating variables, I'm using same-age plants of one variety. They're side by side, so the soil they're in is similar. Each plant got the same amount of fertilizer when I set them. They're receiving the same attentions to watering and weeding. I even drench the control group with the same amount of water the other group receives as tea.
Science is about being as scrupulously honest as we can be. Accordingly, I'll need to observe carefully and report results accurately. Given human hopes, expectations, and gardening biases, these are harder tasks than you might think, but they're a critical part of the experiment.
My test won't make or break the case for tea as a control for late blight. If six of us conduct similar tests this summer and all of us have similar results, we'll be able to draw some tentative conclusions about the effectiveness of tea. And if we or others can duplicate those results next year, we can make the tentative conclusions of the year before still stronger.
