In some cases it is easy to make this distinction. For example, if an investigator has to buy a chemical for a specific experiment, then that clearly is a direct cost to the grant. On the other hand, an investigator's use of electrical power, water and other utilities, or the services of the purchasing and accounting offices, are not normally charged directly because it is not practical to account for them separately. Installing individual meters to monitor usage levels of electricity, and carrying out the associated accounting and billing functions, would probably cost as much as the electricity itself.
Attributing an appropriate F&A cost amount for the use of research space for grant-related activities can be even more difficult. If, as is typical, a building houses dozens of investigators who are involved individually and collectively in teaching, research, public service and other functions, determining the building costs that should be attributed to a particular faculty member's research projects is not practical. For example, each faculty member may have several grants, which may use common space differentially. Although one could imagine a means of attributing a cost for the repair of a section of the roof (which may last 20 to 30 years) to a specific grant, it has generally been agreed that using a more macroscopic and statistically averaged method is much more sensible and cost effective. The basis for distributing space related costs is an annual space study.