In nutrition studies, it is common to compare diets that vary in level of nutrients or in the content of some additive such as antibiotics, enzymes, or anticoccidials. Because differences in anticipated response may be quite small, it is important to assure that the observed differences in response are due to the nutrient or additive in question, and not to mix-to-mix variation in feed ingredient composition.

Even under the best of circumstances, if one were to make 10 consecutive mixes of the same formula, chemical analysis of the diet would indicate minor differences in composition which may or may not be sufficient to cause a difference in performance. In order to minimize these differences, nutritionists may incorporate one of several schemes that we have followed in our laboratory to use in preparation of diets. Depending upon the types of diets being used, one or more of these approaches may be appropriate.

1. Make a great big mix

This approach works well when a common basal diet is to be supplemented with various amounts of some additive. If a mixer of sufficient size is available, one batch of the basal diet is mixed and aliquots used to supplement with the test article. Unfortunately, this approach is usually limited to studies with small numbers of animals, or the starter phase of a larger study, due to the limited capacity of mixers in most research units. Integrated poultry companies typically have mixers with much larger capacity and it is sometimes possible to obtain feed from such producers. However, one must then be concerned about possible cross-contamination with the various growth promoters, anticoccidials, etc that may be in the feeding system of the integrator.

2. Make several batches and blend

This method is effective for mixing diets in which rather large amounts of a common feed are to be supplemented with various amounts of some additive. A number of mixes of the basal diet are prepared and kept segregated after mixing. The final diets are then mixed by taking equal aliquots from the various mixes and blending these together. This procedure is recognized by the US Food and Drug Administration (FDA) as an acceptable means of preparing common basal diets for tests with feed additives.

3. Use least common amount' mixing

This method is effective when a series of diets varying in nutrient composition are to be mixed. For example, the experiment may require a series of diets with different protein content to be compared (see table Composition of test diets'). Although the overall formulas differ, there is a large common component among the various ingredients. In the least common amount' (LCA) approach, the lowest common amount of a particular ingredient is determined, and that amount is used in the LCA diet. When making the final mixes, an appropriate portion of the LCA diet is then blended with the remaining amounts of corn, soybean meal, or other appropriate ingredients (see table Final mixes'). This procedure is very useful when most of the diets are similar, including vitamin and mineral mixes, salt, and other minor components that may not blend as well as larger components.

4. Use a corn-soy-fat blend

One of the common problems in mixing small amounts of diets for research purposes is the addition of small amounts of supplemental fat or other liquids.  This is especially true with certain types of mixers, such as 'V mixers' that are not equipped for liquid additions. To aid in overcoming this problem, one can prepare a mixture of corn, soybean meal, and fat and use this mixture as an ingredient in formulation.  A blend of 75 percent corn, 12 percent fat, and 13 percent soybean meal makes a satisfactory mixture that still retains good 'flow' properties. In formulation, the computer's diet formulation software generally uses this mixture in sufficient amounts to provide the desired level of supplemental fat, and then adds additional corn and soybean meal as needed.

5. Blend high' and low' amounts

Quite often a research project evaluates a series of different levels of some nutrient; for example, added methionine levels ranging from zero to 0.25 in increments of 0.05%. Depending upon the sensitivity of the scales used to weigh the varying amounts of salt, minor variations are sure to be encountered. To help overcome this problem, the researcher can prepare a low' diet containing 0.0% added methionine and a high' diet containing 0.25% added methionine. These diets can be then blended in proper proportions to provide intermediate levels of the desired nutrient.

6. Make a serial dilution of premixes

Quite often a researcher obtains a highly concentrated source of a particular product that is to be included in the diet at very low levels. Weighing and blending a very small portion of this product into a diet can be challenging. In such situations, we prefer to conduct a serial dilution beginning with an amount that can be weighed with some degree of accuracy. For example, we recently conducted a study comparing sources of vitamin D. One source was provided as a concentrated premix containing 500,000 IU per gram. We wanted to compare this at levels of 250, 500, 750, 1000, 2000, and 4000 IU/kg of feed. Obviously, to weigh out directly the amount that would be added to a kilogram of feed would be very challenging. Therefore, we began by adding 10 g of the concentrated premix to 10 kg of finely ground corn meal and then carried out serial dilutions using the 10 kg mix to provide us with final premixes that could be more readily weighed for proper dispersion in the final test diets.

All of these approaches are used in research trials in our laboratory and have been found to be very useful in minimizing batch-to-batch differences in diets and allowing for more precise mixing of minute amounts of ingredients. They can be easily adapted to almost any location where feed mixing equipment is available and can be applied in almost any appropriate situation. Use of these procedures should help the nutritionist in providing for minimum unanticipated variation in nutrient content. FM