Different uses of the term 'replicate'
The dictionary definition of a replicate is "a close or exact copy". In the context of experimental design, the ambiguity is in what is copied.
In completely randomised experiments, the replicates of a treatment are simply the different experimental units to which the treatment is applied. A completely randomised experiment with 5 replicates of each of the 4 treatments is therefore conducted with 20 experimental units.
However when the experimental units are grouped into blocks, the treatments may either be repeated within the same block, or the blocks themselves may be considered to be repeats of experiments with a smaller number of experimental units.
To avoid ambiguity, we recommend only using the term 'replicate' for repeats of a treatment within blocks.
Always be careful with interpreting the term 'replicate' when reading reports.
Rice yields
The table below was published to describes rice yields from an experiment comparing six different seeding rates of the rice. Each seeding rate was repeated in four different plots but the experiment was laid out as four separate completely randomised experiments, each using six adjacent plots.
Grain yield of rice variety IR8 (kg/ha) | ||||
---|---|---|---|---|
Seeding rate (kg seed/ha) |
Rep I | Rep II | Rep III | Rep IV |
25 | 5,113 | 5,298 | 5,307 | 4,678 |
50 | 5,346 | 5,952 | 4,719 | 4,264 |
75 | 5,272 | 5,713 | 5,483 | 4,749 |
100 | 5,164 | 4,831 | 4,986 | 4,410 |
125 | 4,804 | 4,848 | 4,432 | 4,748 |
150 | 5,254 | 4,542 | 4,919 | 4,098 |
Since the six plots in each column were adjacent, a better terminology would be to label the columns 'blocks' and to say that there was a single replicate of each treatment within all blocks.