Four silages harvested in fall 2002 were retrieved from the silo in March 2003 for the sequential water separation. The silages were selected to represent four mechanical harvest treatments: short chop and unproc­essed, long chop and unprocessed, short chop and processed, and long chop and processed. Processing involved crushing and shearing the chopped whole plant through a pair of toothed rolls operating at small clearance and differential speed (10). Silages came from two experimental farms (Arlington, Prairie-du-Sac) and two commercial farms (Binversie, Ziegler) in Wisconsin. A measured mass of 1 kg of fresh silage was placed in a water basin containing initially 7 L of water. After 1 min of manual gentle mixing, the material still floating on the water surface was removed by hand. The rest of the basin contents was poured gently over a screen made of 0.40-mm (1/64-in.) thick wire spaced 1.59 mm (1/16 in.) center to center in a square grid with about 50% open area. The screen separated material into two components, the effluent water and the suspended sol­ids, and a third component was the sunk material that remained at the bottom of the basin after pouring. The latter two components were spread onto separate paper cloths to partially dry in ambient air. The floating material was then deposited again in the water basin with the same efflu­ent water. After 1 min, the floating and suspended solids were set aside for the next water separation and the sunk material was put on a cloth to dry. This process was repeated until eight water separations had been com­pleted. The eight-step sequential separation was replicated three times for each of the four silages.

The sunk material from each of the eight separations, the suspended solids from the first separation, and the residual floating material after the eighth separation (i. e., 10 components) were oven-dried at 103°C for 24 h to estimate the proportions of DM at each step. A well-mixed amount of 2 kg of water effluent was also measured after the eighth separation and oven-dried to estimate the total DM in the effluent.

For each replication (four silages x three replications), the 10 dried components were hand sorted to separate grain from stover. Sorted grain included full and broken grains, grain hull, and grain endosperm pieces that were large enough (1 to 2 mm) to be clearly identified as starch. The rest was considered to be stover. Because sorting occurred over a period of several weeks after oven-drying, rehydration occurred and component masses were corrected to a DM basis. Grain concentration was estimated as the proportion of sunk grain over the total of sunk grain and sunk stover.

Six pretreatments were done to compare the effect of drying or sieving on subsequent grain and stover separation:

1. A fresh untreated silage.

2. Silage that was partially dried until it lost 10 percentage units of moisture.

3. Silage that was partially dried until it lost 20 percentage units of moisture content.

4. Silage that was oven-dried to approx 0% moisture.

5. Silage that was sieved by a standard method (11) and whose material from only screen no. 3 was hydrodynamically separated (particle size between 9.0 and 18.0 mm).

6. The same sieved material as in item 5 that was also partially dried to lose 10 percentage units of moisture.

The silage for all six pretreatments was unprocessed and came from a commercial farm (Manthe) in south-central Wisconsin.

A single water separation was done with these treated silages. Using the same amounts of silage (1 kg) and water (7 L) as in the second experi­ment, the material was separated into three components: sunk, suspended, and floating material. DM in the effluent was estimated by mass balance. The three measured components were further subdivided into grain and stover by hand sorting after oven-drying. The water separation was repli­cated three times for each of the six silage treatments. In the case of sieved material, 1 kg was placed in the separator, and only the fraction retained on screen no. 3 was separated by water.