"Censo!" she shouts, giggling, and her playmates pause in their poolside scamperings to stare at her.

I smile at her, and her look-alike in the red bathing suit. They must be Rosa and Maria, from the downstairs apartment with the open door. I have not met them, but I clearly remember the erasing I had to do a couple of days earlier when their mother, quite by chance, corrected my first impression that Rosa Maria was a single 9-year-old. "No, no," she had said, laughing. "Dos. Rosa. Maria. Tweens!"

I don't know them, but they sure know me by sight, for I have been haunting this apartment complex for a week, the vinyl "U.S. CENSUS" bag full of forms over one shoulder as I check apartment numbers against the entries in my big census log-book, then knock, knock on the metal doors. I know who is sick, who was recently robbed at gunpoint in her apartment, which apartments respond to my knocks only with rock music or yapping dogs.

As you can imagine, rounding up the stragglers in the national headcount is pretty crummy work. Although the census director, on the front of each of the 1980 census forms, describes the endeavor as "a vitally important national activity," the pay that we foot soldiers receive is so low that it insures no very-high degree of accuracy. The task has always been performed by novices, and novices tend to make mistakes. I know I did.

Of course, there is a permanent cadre of professionals, in the Census Bureau, to supervise the whole operation and guard against mistakes. They (more likely, their budgeteers in Congress) have made a colossal mistake if they truly believe they could buy first-rate, short-term clerical help for $3.50 an hour. That may be attractive pay in some parts of America; I've heard that laid-off auto workers were waiting in line outside Detroit's census office. But here in Houston, the fastest-growing of our big cities, $3.50 an hour is not simply laughable, it's an insult.

I don't know whether other census-takers, in Green Bay or Anacostia, would find my experiences familiar or highly typical, but I dare hope Washington might benefit from a perspective from the trenches. The experience was not all negative.

In early April, when the 1.3 million census forms that had been mailed to households here came rolling in, the return rate was running around 60 percent, compared with 75 to 80 elsewhere in the country. That meant more door-to-door enumerators would be needed to tally the people who hadn't bothered to mail in returns.

On Easter Sunday, as I discussed with my parents the numbing solitude that can grip you when you are living alone with a writing project, it occurred to me that it might be good for me to go out a few hours in the late afternoon for a few weeks, knock on doors, meet the people, perform a clearly needed service as a citizen and, yes, maybe a few bucks in the process.

The next day I went to the West Houston census office, a room of cardboard tables and desks in an office complex that looked as if it had been completed the previous Thursday, for a one-hour, 54-question test that wasn't as effortless as I had expected. On Friday, three days later, I got a "welcome" postcard from the Census Bureau, directing me to report Monday morning for two days of paid training, in the Boy Scout room of the Episcopal Church of St. John the Devine, on River Oaks Boulevard.

The church is at the gateway to one of the swankiest neighborhoods in America. My crew leader, Walter Payne, turned out to be a retired gentleman who is an active parishioner at St. John's. Most of us census-takers were males, with an average age I would guess at 33. There were a couple of white college students, a couple of young blacks -- including a saleslady at a jeans store -- but only one full-time housewife.

Right here is where I think Washington has messed up in its conception of the 1980 census. In relying on a vast army of short-term, minimal-wage workers to conduct the task, the Census Bureau has simply not taken into consideration its own reports on the rapid rise of women in the nation's labor force. Those public-spirited housewives, who in past decades could be depended upon to volunteer for census duty with the same zest they brought to the Junior League and the Parent-Teacher Association, still exist, but not very many of them. They are now in the law firms, on the assembly lines, leading the push for better day-care facilities, complaining about the scarcity of good domestic help.

The civic-duty nature of census work still attracts people, despite the absurd pay. One remarkably bright college student in my training class told me she was there because she was "just curious about how it works, and I heard they needed people." To her, it was like an eclipse -- she had been only 9 the last time -- and she could get better-paying work when it was over in a few weeks.

A semi-retired lawyer who was counting his neighbors in River Oaks told me he commanded $60 an hour at his office. "I'm sure not doing this for the money," he snorted. "It's patriotism, pure and simple. We in Houston have got to get everyone counted so we can make our muscle felt in Washington, and not be dictated to by the Northeast on energy policies."

Local pride, civic duty, a bit of the old noblesse oblige, curiosity -- those things still attracted a vastly shrunken cadre of census-takers here (450 of 1,105 projected for the West Houston office), but the reality of the marketplace is a louder drummer. So, I suspect, is fear of crime. Not many retirees or young women were in my class, preparing to knock on doors in strange neighborhoods.

By the end of the second day we had our little badges to wear, our shoulder satchels full of six or seven kinds of forms and notices, our heads spinning with instructions on how to count someone's father-in-law visiting from Duluth if there was no one at home in Duluth to mail in his return on April 1.

We had everything but our assignments, the block-by-block logs being compiled from the mailed-in forms back in that cardboard office. Payne promised me to call me as soon as he had an address log, one close to my neighborhood of untrendy older apartments and middle-class homes.

But it was not until eight days after I walked out of the training, with the census details fresh in mind, that Payne called.

"I gotta book, I gotta book, I gotta book," he says. We meet in the parking lot at St. John's as he arrives for Bible-study class.

Whattta book!

The United States Census of 1980 for District Office 3015, Enumerator District 1268, West Houston, Texas -- a long spiral notebook -- would not inspire confidence from any inspector, even a 9-year-old.

I first study the map of the district, in an envelope in the back of the book.

It is very narrow, mostly commerical silver stretching maybe 2 1/2 miles along the busy Southwest Freeway. According to the map, the lengthy boundaries of the district are the freeway and a street called Westpark. I feel sure Westpark ends halfway across the district. I consult a map. Sure enough. Either four of my census blocks are nonexistent, or have some boundary not noted on this map I have been given.

The map, if vague, is at least neat. The logbook, listing line-by-line each residence in the district, clearly reflects a chaotic census headquarters. e

Virtually all of the 271 places I am to visit are at one address, an apartment project a couple of miles from me. The computer-printed listing of apartments begin erratically: Apts. 1, 10, 11, 113, 114. Alongside each apartment that has returned a census form, someone has neatly penciled the name of the resident, the number of people living at that address, and the date of the entry (yesterday!). About two-thirds of the entries are blank, doors I shall have to knock on.

Following the computer-printed entries, ending in typical fashion with Apts. 86 and 9, are two pages of apartments that have been logged in blue pencil with the uncertain penmanship of a 7-year-old. Whoever wrote these numbers really labored over each gnarled numeral. Curiously, no one at any of these blocks of apartments is recorded as having returned a census form. a

The next 10 pages, all ripped from other logs and inserted loose-leaf, list 200 more apartments in that now-familar 177-241-242/nonpattern. After virtually everyone, someone had written in ballpoint (we were instructed to use only pencil) that it is a duplicate of an earlier entry, please see serial number such-and-such back in the computer-printed section.

My God. Somebody at $3.50 an hour logged in units that were already in the book, then somebody else at $3.50 an hour had to write in notations canceling these duplicates and citing the original serial number for each of them!

I fixed myself a stiff drink and promptly went to sleep.

With misgivings, I walked into the apartment office the next afternoon with my badge and satchel, having gathered from a billboard on the freeway one source of possible confusion: The place was a motor lodge as well as an apartment project.

Scores of apartments had been renumbered and renovated as motel-like "executive suites." Many stood vacant, some with windows smashed out, awaiting conversion.

"One census-taker already came by here a few weeks ago and we went over the whole thing," the manager said, in understandably weary confusion. "Which units are motel, which are apartments, which were vacant. We're all pretty busy, getting ready for a big convention. I guess Sharon can spare a few minutes off and on to help you, but right now she's busy."

I check a map on the office wall and walk out into the sunshine. Against the built-Thursday newness of so much of Houston, this 20-year-old project had a weathered, shabby air. But it had been, and still-was, pleasantly landscaped.

I complete my first form with a young white couple, "now-married," they answer. I soon discover that to be an exceptional status among the English-speaking residents of the apartments, most of whom are young, transient, friendly and often without much furniture. I get used to conducting interviews seated on shag carpeting.

That first day I handed out four forms to residents who were so uncertain in their English that my uncertain struggles with nombre (name) and navidad (birthdate) were useless. Maybe navadad means Christmas? The forms -- copies of which, with their names and addresses, go to the census office -- tell them in Spanish that a Spanish-speaking enumerator will visit them later. I think.

My first interview with a long form (for which we get top piece-work of $3.80) does indeed take the 40 minutes they said it would in training -- and with a household of only one person, a young black man from Louisana who describes his ancestry as "Creole." I've never been sure exactly what that meant, but he says it after some reflection, so I letter it in without comment.

Similarly, another young man a few doors down spends several minutes describing whether to list his race as "white" or "American Indian." His statement that he is half-Cherokee surprises the girl who shares the apartment with him and her boyfriend.

"Well," he finally says, "I guess 'white' is okay. Just put down 'white.'"

The seperate race and ancestry questions are among the most troublesome on the census forms for many Americans, myself included. I'd be hard-pressed to sort out an ancestry that's pretty evenly divided between the British Isles and Germany. Despite the Irish name, my Delaney ancestors have been Protestant Americans for more than a century and regarded March 17 as just another workday. I guess I would put English as the root of my mongrelization that I feel most comfortable in identifying with -- and that's what the Census Bureau wants. When a person mentions two or more ancestries, we are instructed to listen for and record the one mentioned first.

The Hispanic-ancestry question, which subdivides into Mexican-Puerto Rican-Cuban-other and surely will be on our census forms for many, many decades, led one young mother to describe herself as Mexican, her husband as Cuban, and her children, proudly, as "Americans." I list the ninos as "Hispanic-other," as I do a woman who was born and reared in Spain.

I do a long form on a young engineer from Lebanon. He gets out photographs of the heat-exchange units he designs for petrochemical plants as we struggle to translate his occupation into "tricky censusese ("engineer," "doctor," "teacher" and "nun" are regarded by the Census Bureau as unacceptably vague).

The people who answered my knock were almost unfailingly coopertive, even hospitable. Three young men from Lebanon, wearing pajama bottoms in the middle of the afternoon, courteously offered me tea. I did accept Tab later from an even younger group of immigrants from Michigan.

People were very precise in reporting their incomes last year, undoubtedly because their memories had just been refreshed by the income tax filing deadline. Several dug into closets and desks for copies of their 1040s and W2s. "$376," announced a 17-year-old boy fresh out of high school.

A census-taker finds a much more wonderfully open and honest and trusting people than he dares to expect.

Census-taking, I quickly learned, is incredibly seat-of-the-pants work, relying heavily on the enumerator's common sense, grasp of the rulebook, a cool display of respect and professionalism that must be obvious to the interviewee, and, as with everything else, sheer luck.

The enumerator exercises considerable latitude, and there is ample opportunity for cheating. I mean, after five futile calls to an occupied apartment, wouldn't it be just as easy to mark it down as vacant, collect $1.75, and forget it?

Of course, I wouldn't do that, but I have done some things that probably weren't kosher.

A man in his thirties told me he and his roommate -- a much younger man -- had just moved in two weeks before and thus weren't at that address on Census Day, April 1. Were you counted where you were living before? No, I've just been kind of drifting. I count him as a resident of the apartment on April 1.

The roommate says he's "pretty sure" his parents counted him on their form. The two begin a sort of arch conversation. I know there's some special form or box that I'm supposed to do on the younger man, but it has been two weeks since training and I simply don't feel like digging into my manual and studying it all over again while listening to their byplay. Hoping that the younger man's parents did count him, I gladly made no note of him and leave.

What do I do about these vacant apartments, many with smashed windows, some without refrigerators and stoves. Do I cancel them out as usable housing, for which I get nothing, or do "vacant" forms on them, at $1.75 per form?

I cancel a dozen or so that are in really bad shape -- no doors, for example, or a caved-in ceiling. The others I chose to regard as habitable and merely "vacant." Surely a more aggressive management, with a good line of credit, could repair the windows, have kitchen appliances installed and get them on the market in a few days. It wasn't a very tidy call, but I'm comfortable (and maybe $40 richer) with it. And the management itself describes those units as "vacant."

Census-taking can also be pretty boring.

While I was interviewing one woman, who was wearing only a slip and an olive-green workshirt, the phone rang. It was her mother, calling from Mexico. For the next 10 minutes her 2-year-old daughter and I amused ourselves by making funny faces at one another.

The "bonanza" of the vacant units was more than offset by such waits, by people who said they would be home at 6:30 and weren't, by trips to the management to determine if Apt. 82 had indeed been renumbered 1004, by trips to Payne's apartment to have him inspect my completed forms.

As the holes in my ragged logbook for that project filled up, I discovered new ones elsewhere in my district -- four major motels which had been visited by an earlier census-taker who has properly listed the few units occupied by night managers as permanent dwellings. But the motels themselves had not been logged in the page of "special places" (prisons, hospitals, hotels).

Between two of the motels I came upon a 54-unit condominium that was also missing from my book. It was eerily, almost wholly, unoccupied. Sloppy work by my precursors meant gravy for me, right? Ten minutes with the management, a couple of hours filling "vacant" forms for 40-some units, adds up to 70 bucks.

The management office was locked. I recorded names off the few mailboxes that had them, and began knocking. Men at two apartments greeted me with polite exasperation, saying they had received census forms at that address in the mail and had completed and returned them.

So somebody at the Census Bureau knew there were residents at this address, or they would not have gotten forms in the mail. How come my book shows nothing at this address?

At the third door an attractive young woman in an intriguingly blowsy housedress says that she, too, had received a form in the mail and returned it right away. I believe her, too. But instead of thanking her and moving on, as I had done with the men (and I see this now only as I write it), I politely insist that I do another form on her, just in case. It won't take a minute. She agrees, bolstering my ego with the hint that I may be more interesting than the TV weatherman who had been entertaining her.

She had signed to buy the condo four years ago, she explained, and the developers encountered financial problems. The status of the units' ownership clouded into the courtrooms, sales were halted, and now she was paying rent to a management of which she was only dimly aware. It wasn't a romantic conversation, against the reports of thunderstorms in East Texas, but I copied her name and phone number off the census form when I got home. Well, I may need to call her if I have trouble locating a manager.

A couple of days later, Payne and I discover that the hapless condo is not missing after all. Each of its units is duly listed in a second logbook for my area, to which we both should have noticed references. Payne decently shoulders the burden of correcting my duplicate entries.

In 10 days of enumerating (gunky Washington word -- I prefer census-taking), I have probably put 70 hours of work and earned around $200 gross.

I can be proud that I added about 100 Houstonians to the census rolls who weren't there before (including several duplicates for people who say they mailed in returns but weren't listed in my book).

I am proud, too, that the mixed-up redundant, childishly scrawled pages that dominated my book when I got it now list with reasonable accuracy the dwellings in that area.

But it depresses me to have spent so much time, for which I will never be paid, crossing out scores of hotel rooms that someone else, getting paid by the hour, laboriously and mistakenly logged as apartments -- and logging in myself, mistakenly, a condo I thought was missing. Novices like me will always make mistakes.

Had I reported to the census office at $3.50 an hour, I would have grossed $45 more for my hours worked.

There will always be a few Walter Paynes and curious, mildly civic-minded souls like myself to sustain the illusion that a careful census can be conducted for a pittance of what private industry would charge -- and pay workers -- for such a task. It can't.

The census simply isn't a very vital national priority. If it were, we census workers would be laughing our way to the banks with all those Golden Fleece award-winners.

"We census workers." Gad.

Count me out.