are based on national characterizations of solid waste generation that involve the questionable validity of using government and industry records of production, together with an untold number of untested assumptions, to estimate residential and other discards. Even if the national estimates are accurate, they are available only at the level of categories of material composition—so much plastic, glass, aluminum, paper, steel, and so on. But how can anyone plan on the basis of such data? Most of the materials come from the packaging that people buy at stores, but no one goes shopping for five ounces of glass, three ounces of paperboard boxes, and eight ounces of aluminum cans. Instead, they shop for a jar of Best Foods mayonnaise, a box of Cap’n Crunch cereal, and a twelve-pack of Bud Light beer. This brings the project back full circle to its item-by-item regular sorts of fresh refuse. In other words, in contrast to virtually all other sources of information, the Garbage Project looks at refuse the way all archaeologists do—as the material result of human behavior.

Ultimately, the contribution of the Garbage Project comes down to one simple component: in order to understand and mitigate important problems, we have to, first, become aware of the problems and, second, measure their impact. This difference in approach can mean a considerable difference in results.

Since 1987, communities everywhere have been promoting recycling, reuse, source reduction, and everything else they can to decrease the amount of refuse being discarded. At the same time, to cut collection costs and reduce worker injuries, many communities have converted to automated systems that depend on standard-sized garbage containers. The containers that most families bought for themselves were usually sixty-gallon containers, about what one person could carry a short distance. The new standardized containers have wheels and can hold ninety gallons, one-third bigger, to accommodate the needs of the largest families. By all accounts, the results of these changes have been that recycling is increasing and on-the-job injuries are down.

The Garbage Project’s hands-on sorting, however, adds another dimension—a darker side that is rarely mentioned (Rathje 1993). When the Garbage Project first studied residential refuse in the city of Phoenix, that city, unlike Tucson, already had an automated system, and Garbage Project personnel were surprised to discover that Phoenix households discarded nearly double the refuse thrown out by households less than 100 miles away in Tucson. The mystery was greatly clarified when Tucson switched to the automated system and its household refuse generation rate increased by more than one-third. At this point, the Garbage Project identified a “Parkinson’s Law of Garbage” with implications for every city’s solid waste management strategy (Rathje 1993).

The original Parkinson’s Law was formulated in 1957 by C. Northcote Parkinson, a British bureaucrat who concluded, “Work expands to fill the time available for its completion.” Parkinson’s Law of Garbage similarly states, “Garbage expands to fill the receptacles available for its containment.”

Parkinson’s Law of Garbage is really quite simple. When people have small garbage cans, larger items—old cans of paint, broken furniture perpetually awaiting repair, bags of old clothing—do not typically get thrown away. Rather, these materials sit in basements and garages, often until a residence changes hands. But when homeowners are provided with plastic minidumpsters, they are presented with a new option. Before long, what was once an instinctive “I’ll stick this in the cellar” becomes an equally instinctive “I’ll bet this will fit in the can.”

The Garbage Project has compared the components of Tucson residential refuse collected before and after mechanization. Solid waste discards went from an average of less than fourteen pounds per biweekly pickup to an average of more than twenty-three pounds. The largest increase was in the yard waste category, followed by “other” (broken odds and ends), food waste, newspapers, and textiles. The first pickup of the week was substantially heavier than the second, reflecting the accomplishment of weekend chores, and the discards in that first pickup were loaded with consistently larger quantities of hazardous waste than the Garbage Project had come to expect in a typical load. These findings