A pot experiment in a greenhouse was performed. Two soils were used; a sand (Elverum) and a sandy loam (0ksna) (Table 3.1). The soils were selected for this experiment because of the low content of readily available K, Mg and P which was found by Jeng et al. (2006). Although the soils were sampled at the same locations as the soils used in the experiments of Jeng et al. (2006), assuming low concentra­tions of readily available P and K, analyses of the soils selected for this experiment showed higher concentrations of readily available K (Table 3.2) than were found in the previous investigation [Elverum 7.6 mg readily available K (100 cm3)-1, 0ksna 8.6 mg readily available K (100 cm3)-1]. According to 0gaard et al. (2002), no yield response to applied K can be expected if the concentration of readily available K exceeds 10 mg (100 cm3)-1 (8 mg 100 g-1 and bulk density 1.25 mg m-3).

The experiment was designed to supply NPK similar to that supplied by use of the compound mineral NPK fertilizer Yara Fullgj0dsel® 21-4-10. On the basis of the contents of N and P in MBM, and of K in different crushed rock powders and BWA (Table 3.3), the amounts of different components in the experimental design (Table 3.4) were calculated using the following assumptions for calculation of effective amounts of NPK:

1. The N effect of MBM was estimated as 80% of Kjeldahl N as equal to mineral N based on Jeng et al. (2004).

2. The P effect of MBM was estimated as 50% of total P equal to the effect of mineral P (Jeng et al. 2006).

3. The amount of K extracted by 1 M HNO3 from crushed rock powder and by 7 M HNO3 from BWA was estimated as plant available and equal to mineral K.

Although the total amount of NPK applied differed between the treatments, based on the assumptions for effective amounts of N, P and K, it was intended to obtain almost the same effect of NPK as mineral NPK (treatments 3-12).

The pot size was 7.5 l, and the height of the soil in the pots was 20 cm. The fertilizer level was based on normal fertilization recommendations for cereals in Norway (120 kg N ha-1), and the concentration of plant nutrients in the pots should be at the same level as in a 20-cm-deep plough layer. All amounts of added fertilizer were calculated on a hectare to pot basis. There were three replicates. The crops used in the experiment were spring barley (Hordeum vulgare cv Kinnan) and spring wheat (Triticum aestivum), and the amounts of K applied (60 kg K ha-1) were in line with normal fertilization recommendations for cereals in Norway. Seeding was performed in May with 30 seeds in each pot. After germination the weakest plants were removed, leaving 20 plants per pot. Unfortunately, barley seeds from two different batches were used, and uneven germination of barley was recorded owing to poor germination of the seeds from one batch. However, in the pots with fewer than 20 plants, the lack of plants was partly compensated for by an increased number of tillers. A mean of 19 ears per pot was found at harvest for both barley and wheat. Therefore, the effect of uneven germination was found not to have a significant influence on the yield.

The intended temperature in the greenhouse was 15°C at night and 20°C during the day, but on warm days with outdoor temperature above 20°C the temperature inside the greenhouse was somewhat higher than the outdoor temperature, reaching 30°C during some summer days. The pots were initially irrigated 3 days a week, but in

warm periods irrigation was carried out almost daily to prevent drought. The irrigation caused no leaching of plant nutrients from the pots during the experiment.

The nutrient balances were calculated on the basis of effective amounts of P and K applied and uptake of P and K in wheat grain. In the experiment straw was not harvested and nutrient uptake in the straw is therefore not included in the calcu­lations of nutrient balance. After the experiment had finished, soil samples from all treatments from both the wheat and the barley experiments with both soils were taken (0-20 cm), 48 samples in total. The samples consisted of nine subsamples, three from each pot. The results for the soil samples in Table 3.7 represent means for both soils and both crops.

3.2.2 Statistical Analysis

One-way analysis of variance was carried out. For multiple comparisons the Ryan- Einot-Gabriel-Welch (REGWQ) multiple range test was applied with a signifi­cance level of P = 0.05, and the means presented followed by the same letter are not statistically different.