When workers for this year’s U.S. Census approached Fraser Bresnahan of the Coalition for the Homeless for help counting the poor and homeless, he was eager to help. Bresnahan oversees two vans that serve 700 meals a day at 25 locations across the city, and he readily provided the Census Bureau with the vans’ schedules.

But census workers made it to just four of the 25 sites Bresnahan had suggested they visit. And once they got there, the enumerators dawdled so long that many of the diners left before they could be counted. “They stood around and waited for us to begin serving before they began counting,” Bresnahan says in disbelief. “Obviously, once people get their food they want to eat and leave…. They were hugely disorganized.”

Clyde Kummerle, whose Holy Apostles Soup Kitchen on Ninth Avenue is the city’s largest, had his own chaotic encounter with census takers. When the workers arrived, they didn’t even have pencils, never mind a clear idea of how they’d interview a sampling of the 1,100 people who pass through the soup kitchen for lunch. Census crew leaders, who might have had a clearer idea of how to conduct a count of homeless people, were nowhere to be found. “No one really had a handle on what to do,” says Kummerle. “I believe one of our staff or volunteers went out and got them pencils.”

Ten years ago, the Census undercounted New York’s homeless by more than 70,000. Overall, the 1990 count failed to enumerate 450,000 New Yorkers, costing the city an estimated $450 million in federal funds. An undercount affects the city in many ways. But a low count hits the neediest the hardest, because much federal funding for social programs is based on the proportion of poor people in a given area. According to the finance division of the City Council, the 1990 undercount cost the city $94.5 million in community development block grant dollars, $42 million for Head Start, and $3.9 million for emergency shelters.

But this year’s losses may make 1990 look like a windfall, according to a report drawn up for the federal Census Monitoring Board by the accounting firm PriceWaterhouseCoopers. In the event of a serious undercount, the study forecasts a steep decline in federal assistance for adoption, foster care, and vocational education, among other programs. The total price tag is an estimated $2.3 billion.

The repercussions of the 1990 undercount of the homeless were severe, as conservative lawmakers used the low figures to justify cuts in funding for programs for the homeless. As a result, advocates for the homeless helped convince the Census Bureau not to issue a figure this time around measuring the total homeless population.

But a failure to count in homeless people still distorts the Census’ overall picture of the number of poor people in New York City. To make sure that the census’ long form, which measures income levels, got to the appropriate number of homeless people, the Census Bureau agreed to stage a two-day count at the city’s homeless shelters and soup kitchens.

“For our purposes, the most important aspect of the census is the chance it provides to accurately count [the number of people] living in poverty,” says Mary Anne Gleason, a housing policy analyst for the National Coalition for the Homeless. Gleason also advised the Census Bureau this year.

But this year’s count isn’t cutting it, say census workers. Citing the confidentiality of its clients, the city Department of Homeless Services refused census workers entry into all city-run shelters. According to one enumerator, DHS told them that it had “made its own arrangement” with the Census Bureau. In the deal, DHS collected its own data on homeless people on the census short form, which does not note income. As a result, the Census collected no information on the income of many of the poorest people in the city. (Cenus officials declined to comment on the New York count.)

Gleason says the impact of the shelter undercount is significant. In the shelters, “New York has about 25,000 homeless people on any particular night,” Gleason explains. “If this entire population had been polled and the appropriate percentage given access to the long form, potentially more than 4,000 additional names would have contributed to New York’s poverty figure.”

Census workers confirm that efforts to count the poor and homeless who don’t live in shelters were just as careless. Enumerator Frank Eadie says he was sent to a soup kitchen in lower Manhattan that served Chinese clients, and instructed to count and interview them. Speaking no Chinese, Eadie could do nothing.

“I’d gone to work for the Census Bureau in hope of improving the inclusion of the homeless and hard-to-reach population,” says Eadie. “But my commitment, energy and time proved wasted.”

Additional reporting by Judy Richheimer.