Analys av ett potatisförsök med olika parcellstorlekar

This report is based on one field trial with different plot sizes in potatoes. The trial was carried out in Röbäcksdalen, Umeå during the years 1983 to 1985. Lack of randomization makes the result less accurate for generalized conclusions. The report describes how the variation between plots varied depending on plot size. Plot sizes in the trial was of the length 7 and 15 meters combined with the width of one to five rows. The inter-row space was close to 70 centimeters. Larger "parcels" in the report consists of aggregations. The trial confirms that larger parcels gives more information per parcel and that the information per unit field area increases by decreasing plot size. The trial also shows that the inter-row competition is an important factor to consider when choosing convenient plot size. Nothing in the report can be regarded as new knowledge but the report gives some examples of important factors that role the chose of suitable plot size. With the limitation to tested plot sizes the net plot size of length 7 meters and 2 rows width are recommended. An important recommendation is that the whole plot include one border-row on each side. Single row plot sizes should be avoided even if border rows are used. The competition between neighbouring rows gives an extra high variation between single rows that are “filtered" away by choosing double rows. An other argument for not choosing single-row parcels is the fact that the ratio net area versus field area becomes rather small when using border rows. Increasing the parcel width to 4 rows or more rows does not give more information per unit field area. Larger parcels also makes it more uncertain to take representative samples for further analysis. Even number of rows is better than odd number depending on the inter-row competition witch can result in an "every second row effect". This report can also be regarded as a small collection of statistical methods usable in plot size trials. For further studies of most convenient plot size it is unavoidable to study the effect of competition between neighbouring rows (border effects). To estimate most convenient plot size depending on soil heterogeneity is possible from the point of known theories. But the value of such an estimation stands and falls with the bordering effect when the plots become relatively small