Beyond those cost items that are physically, contractually, or empirically traceable, some degree of arbitrariness enters any attempt to identify additional costs for a product unit. For example, the traceable material and labor costs of a small number of defective units that may be produced along with the good units could be included logically as part of the cost of the good units. But exactly how much should be included is subject to debate. Is it the average actual amount of the cost of defective units? Is it the actual amount that occurs on the next production run? Or is it the amount that would occur under ideal conditions (which may be zero)? Even when the purpose of the ultimate cost measure is known in advance, the answer to this question is not always clear.