DR. PREWITT: Good. Thank you. I suspect that some of you do have a special interest in a story that was reported late last week. However, since this is our standard operational press briefing, and in fairness to those of you who came with that expectation, I do want to detail a few of the major operations other than talk about the advance letter.

First, update/leave. Our first major labor-intensive operation is ready to begin on schedule this Friday, March 3rd. It runs through March 30th. In this operation, we hand-deliver the questionnaires to about 20 million housing units, mostly in rural areas. We have hired and trained 73,000 temporary workers for this job. They will update address lists and maps as they go, and they will leave the questionnaire package with someone in the household. If no one is home, they leave the package in a plastic bag on a door knob.

Recruitment. As you know, the census is truly a national mobilization effort. We require a huge labor pool, a labor workforce of approximately half a million workers to fill the 860,000 different short-term, local census jobs. As of last week, we continue to be ahead of schedule in this recruitment drive. Our overall recruiting goal is to interest 3 million applicants to reach our hiring -- in order to reach our hiring goal we need to interest 3 million applicants. By last week, we had tested over 2 million applicants.

Now, not all applicants qualify. The most important benchmark is whether we're attracting qualified applicants. In this respect, we're currently 4 percentage points ahead of our schedule. Last week, our qualified applicant pool reached 1.7 million, which is approximately 70 percent of our objective, 69 percent of our objective, and our current target has it at about 65 percent. So that's to reach a total of 2.4 million people. So we have a qualified applicant pool now of 1.7 million people.

Census in Schools. All schools K through 12 have been invited to participate in teaching about Census 2000 through the Census in Schools project. This has been one of our most successful and innovative promotional efforts. By last week, we had shipped over 1.2 million teaching kits containing lesson plans, wall maps and take-home materials to schools around the nation. In addition, numerous members of Congress, the Cabinet, other federal officials are taking the message of Census 2000 to students in their classroom. One example, last week here in Washington, D.C., at the Bruce-Monroe Elementary School, Commerce Secretary William Daley, White House Chief of Staff John Podesta, D.C. Mayor Anthony Williams and Delegate Eleanor Holmes Norton met with the school children.

I pause to say that, in addition, that event was attended by the Count, Sesame Street's Muppet character, and my reports are that the Count was of more interest to the kids than the Secretary of Commerce, the Mayor of Washington, D.C., the representative and the chief of staff. Nevertheless, it was a very successful event, and we are extremely pleased that the Count was able to join us and stress the importance of counting.

The next big step in our Census in the Schools program is Monday, March 13th, that starts "Teach Census Week," where we hope every classroom in America will have census lessons taught.

Road Tour. In another innovation we have commissioned 12 rolling Census 2000 promotional vehicles, large vans especially outfitted to carry the message that the census is coming. We call this the Census 2000 Road Tour. The national road tour kickoff came February 16th, live on NBC's Today Show. Other appearances have been in Los Angeles at the Grammy Music Awards; Chicago, an event attended by the Rev. Jesse Jackson, and the former boxer Rubin "Hurricane" Carter. A large number of such events. Indeed for the next six weeks the 12 vans will be on the road, enabling our staff to reach tens of thousands of people at stops along the way -- shopping malls, schools, ethnic festivals, parades. I believe that a large number of the vans, for example, will be in St. Patrick's Day parades, community centers. And in just one week of the van tour, we had already made 190 stops and generated coverage in 170 media outlets along the routes that the vans are covering. More than 75,000 people have seen the vans in their communities.

Finally, before we turn to the advance letter, How America Knows What America Needs. This, you know, is the challenge to elicit full cooperation by the American public and to make Census 2000 into a unique civic ceremony, one which does indeed celebrate our diversity and summarizes our needs as a nation. We must present the census to every household in this country as more than a statistical head count. We must show through our advertising and promotional efforts, partnerships that the census is in fact the measure by which America does know what America needs.

For every community in the country, therefore, we have set the goal of trying to increase the response rate by 5 percentage points over the 1990 base. We've called this program '90 Plus five. By last week, we had enlisted more than 2,600 participants in this effort, including a large number of the largest cities -- Houston, Dallas, Los Angeles, Atlanta, Detroit, Washington, D.C., Philadelphia, San Francisco, Baltimore, Fort Worth and so forth. State governments are also joining -- California, Ohio, Michigan, North Carolina, Mississippi, Nevada, Oklahoma, Wyoming and others.

Starting March 27th, we will post response rates for every locality on the Internet, and I think also in the press. And, these rates, of course, will be updated daily through April 11th. If we succeed in this program, we will raise the nation's total Census 2000 response rate to 70 percent -- five points higher than 1990. That will mean millions of fewer houses that we'll need to check by going door to door in the Non-Response Follow-Up period.

Now I turn to the advance letter mail-out. As you know from these operational briefings, we usually try to alert you to the next major activity and describe that in more detail. We had long ago planned to discuss what we call the advance letter, since our schedule calls for it to begin arriving in American mailboxes this coming week. The advance letter is a new feature in the 2000 Census, the result of rigorous testing during the last decade. It is intended to promote awareness that the census questionnaires are on their way. And, in fact, it appears that we've already generated some awareness of the advance letter.

Indeed, we were meeting just a week ago, planning this particular press briefing which, as you know, was initially scheduled for tomorrow and then was moved to today. And we were talking about how could we get the press to pay attention to the advance letter. We did.

During the next two weeks the Postal Service will deliver approximately 115 million advance letters mailed first class, alerting people that the Census 2000 will soon be arriving at their homes -- Census 2000 questionnaires. This letter is the first of three deliveries from the Census Bureau. The others are the questionnaire package itself and a reminder postcard. Let me say up front that we already know that the Postal Service will be unable to deliver 5-to-7 percent of the letters as they are currently addressed. The reason is that we have created our master address file by being as inclusive as possible. We have solicited address information from all possible sources. We have field-checked every address.

However, when an address remains uncertain, we have given it the benefit of the doubt to ensure that it receives all possible chances of enumeration. In addition, the USPS is delivering the advance letter in rural areas. Rural addresses are not compatible with the Census 2000 geographic processing system. We cannot automaticall assign a rural address to a block face. Therefore, we developed the address list and deliver the questionnaires in these areas using census enumerators.

Nevertheless, in order to raise census awareness in those areas, we've attempted to send an advance letter when we believe we have a mailing address. Past experience shows that a portion of those addresses for the rural areas will not be consistent with those that the USPS uses for its delivery.

Now, on Tuesday we learned that there is a problem in the printing of the house number on the address line of the advance letter. It is not a problem with the address lists themselves. You will ask, how did this happen and how certain are you that it will not happen again. I will address both questions, but first put the problem in a larger context. The problem is cosmetic rather than operational. In describing the problem as cosmetic, I do not intend to trivialize it. The public image of the census is nearly as important as the actual operations of the census and, indeed, the image can affect operations because the census does finally depend upon public cooperation.

The advance letters will be delivered by the USPS to the correct households. Moreover, the address mistake does not affect any other census operation. Whereas the bar code in the advance letter does connect to other operations, the household address line does not. And absolutely in no way does the mistake put Census 2000 at risk, and in no way does it affect the census count for 2000.

Here, then, is our key message to the public. When the letter arrives, the American people should open it, read it, not discard it and not try to return it. Starting next Wednesday, March 1st, the first batch of these advance letters will arrive in rural areas. About 17 million households, those with rural-type addresses, will receive this letter.

The second batch of advance letters, addressed to households in areas with what we call city-style addresses, will begin to be delivered Monday, March 6th. About 98 million households will receive this letter. The crucial moment for Census 2000 only comes about one week later. That's when households will receive the correct questionnaire to fill out and return by mail to the Bureau.

This advance letter is simply a heads-up to people on why this census is important. We think that when people are informed, they will make the decision to fill out the form and send it back. Our research indicates that this heads-up will improve response rate by 1-to-2 percent. Equally important, the advance letter provides an opportunity for households which need a form in any one of the five languages other than English to request one. That is, in Spanish, Korean, Chinese, Vietnamese or Tagalog.

As you can see from the displayed example, the bottom of the letter -- this is the display of the advance letter. The bottom of the letter has five instructions in language. It's simply -- these five sentences simply say, turn the letter over. And when you turn the letter over, this is the back of the letter in the five languages. What each of these messages say is that if you wish a form in one of these languages, simply check the box and mail it back to us. And there is, of course, in the envelope an enclosed postage-paid return envelope.

We will take special steps to mobilize nearly 6,000 local, regional and national organizations who are already our partners and who serve the concerned population groups, especially the linguistically isolated groups. Together with our partners, we will mount a comprehensive outreach effort, including letters, telephone calls, fliers and posters, as well as targeted paid advertising. Our message will stress that the letter is important and that it was delivered to the correct address. We will urge every household to open and read the advance letter because that's the only way to request the foreign-language form, which must be especially coded for that specific household address.

We are confident in the effectiveness of both our message and the strategy to implement it. The GPO is investigating exactly how the error occurred. The error, as I've stressed, does not affect the accuracy of the address in the eyes of the Postal Service and its letter carriers. That address and the USPS bar-coded form with ZIP plus four is correct. The household address is also correct on the Census Bureau's separate bar-coded information for that address.

We know that the right advance letter will get to the right address. So does the Postal Service, and so should the person who gets it in the mail. The reason? The codes used to sort the letter for this letter carrier are not affected by this anomaly. The automated sorting equipment will function correctly and place the letter where it ought to go in the rack for the correct order of mail delivery.

Second, the Postal Service has alerted all of its local post offices to be sure that the letter carriers are aware of the problem in the address line. The Census Bureau is grateful to the Postal Service for its rapid, its efficient response on this issue and, of course, for its reassurance that the formatting error will not affect delivery.

Bottom line: the mail will go to the right addresses. Every recipient should open and read the letter. Most recipients won't need a foreign language form so they have no need to respond to this letter at all. Those who do, can use it to request the language form. The system in place will work.

Having described the mistake, I do wish to emphasize that the Census Bureau's quality control procedures failed to detect it. While we cannot accept responsibility for the mistake itself, we accept full responsibility for our failure to detect it. Incidentally, one press report suggested that I blamed the printer. This is not entirely correct. I described as best I could what happened, but blame is not a word I used or would use. I say this not to quibble with the press report, which was accurate in all important respects, but only to emphasize that the responsibility rests with our quality-control procedures. On this particular item they were inadequate.

I've instructed staff to rescrub every one of our 24 major quality control procedures. That review is underway. Is it possible that an error of this sort will happen again? Though anything is possible, the odds are so low that my honest advice to you is to look elsewhere for your next census story.

If not an error of this sort, will something else go wrong in the census? Probably. The census is a vast, multi-part, rapidly moving system with hundreds of operations and hundreds of thousands of recently hired employees. Something could go wrong. Whenever and whatever that something is, it is my confident expectation that it will not put the census itself at risk. At the present time, based on what we know about the robustness of our operations, the only thing that can put the census at risk, other than a natural disaster, is a significantly lower response rate than we are expecting.

When the census form arrives, the first major task for the Census Bureau will actually be complete. When the form arrives, the responsibility shifts from the government to the people, and this is, of course, the moment of truth. The most simple act, taking the pen -- as I've said, this simple technology, and filling out the form is what we will hope the American people will be doing in approximately two-to-three weeks.

With that I will take your questions.

MR. JOST: If I may, we've got a system here, for those of you who are new. We have reporters on the phone, and we'll try to alternate between the room and those on the phone and, therefore, we ask everyone to identify themselves and give their affiliation.

Yes, in the front row.

Q: Dawn Holly, NBC News: Can you name the contractor? Can you tell us which state and city the contractor is from?

DR. PREWITT: In your press kit there is a GPO statement that identifies the contractor.

I will refer all questions about the vendor itself to the GPO, which is doing the investigation.

Q: Haya El Nasser, USA Today. Is the same printer handling the address labeling for all of the mass mailings, including the questionnaires and the follow-up thank you/reminder letter?

DR. PREWITT: No. We have a large number of printers, and this printer is not handling the follow-up letter. All of the census forms themselves have already been printed. This printer had one part of that relatively small part of the larger printing job for the census forms themselves. As you know, we've talked about that. I think there are 85 different forms. We've reported to you the total print run is something like 420 million forms.

We obviously have spent a lot of time over the last 3 days not only checking that printer's work but every other printer's work. We have re-examined every one of our runs, making sure the bar code and every feature of the address is correct. And we are absolutely confident that there are no other errors in the print runs associated with the census form itself, or with the postcard or with the language form that will be sent back if you send this in.

It's very important that I stress that if someone checks this box and sends the letter in, the form that they get back will not have that extra digit on it. That is, a completely separate operation. This was one of the print runs which was unlike a lot of them, as a matter of fact. But this one is completely independent of any other kinds of operations. There are no hooks or links. That is, the address has no other hooks or links to any other kind of census operation.

The bar code does, on this letter. The bar code connects back to getting a form in language, and the bar code is correct.

Q: D'Cohn, Washington Post. Could you talk a little bit more about what went wrong with the quality control here, and whether -- is that a unique lack of quality control? Do you -- in all your other printings, do you make sure that when it leaves the printer and heads for where it's going that there are actual physical checks of the documents, which I understand wasn't done in this case?

DR. PREWITT: Surely. The quality control procedures -- we have lots of different kinds of quality control procedures, of course. But specifically to this instance, we -- after we finished spec-ing out, giving the specifications to the printer for printing this form and its address -- we ran a test deck. And when we ran that test deck on this print run, it tested out perfectly. No errors, including the error that subsequently occurred.

Then we were still on this letter. We were still working with our advisory groups to make sure that the language that was used was exactly the way our advisory groups wanted it to be. After we had run our test deck we then went back to the letter, not to the address file, but only to the letter itself, the wording on the letter, and requested that the printer make some changes.

What that meant, of course, is the printer re-opened the software that drives this process in order to make the changes on the languages. In none of those transactions did we presume, or of course instruct the printer to make any kind of change to the address file itself. Nevertheless, when the software got opened, we presume -- we presume that's when the error occurred, though we don't even know that. The GPO will be doing this investigation.

Now, after what happens, we don't stop our quality control procedures, of course, with just the initial test deck. When we're doing a print run of anything in bulk, we pull off approximately 200 of the print run every four hours, all day long, seven days a week. We take those batches of the print run and send them to Jeffersonville, where they then are subjected to rigorous quality control procedures. The quality control procedures that we are now using on this print run focused, in this instance, on the things which had operational consequences. We were very concerned throughout this entire print run that the bar code -- our bar code appears on this letter twice. It appears -- because it's going to be sent back. But on both the front and the back the bar code appears.

I'm wrong. Our bar code appears at the bottom of the letter and we had to make sure in our quality assurance that the bar codes matched. This is the -- sorry. This set of numbers above the address file, which are our numbers, our identifying number for this residence, and the bar code matched to make this letter work. Our quality control procedures at that stage focused upon that issue because that's what had operational consequences.

We consider this a mistake in our quality control procedures. It did not send our quality control people back to look at an address -- because it was not operationally consequential.

Now, to finish with your question, in terms of other kinds of things. We now -- have 34 major quality operation control procedures. We are re-scrubbing every one of those, which is to say all of them have been designed to make sure nothing goes wrong operationally. We're now re-looking at them to make sure that nothing can go wrong outside of the boundaries simply of the operations itself, which could nevertheless be, as I say, be in this case a cosmetic or -- a public embarrassment to the Census Bureau. So we are now looking at that from that point of view, which we had not done in this instance.

Q: Judi Hasson, Federal Computer Week: You had another test last week of the scanning system. How did that go?

DR. PREWITT: That was the data scanning system. The foresight test. I haven't heard anything, which means it went all right. All indications today are that it went fine.

Q: Herbert Sample, Sacramento Bee. Did you say that 5-to-7 percent of these advance letters will not be delivered by the Post Office? Can you elaborate on that, and then I have a follow-up.

DR. PREWITT: Surely. The address file was, as you know, in the city-style areas where we actually do use the United States Postal Service to deliver the questionnaires. We worked actively with local governments back and forth in our local update file work, the LUCA program. And in that process there are certain addresses which we actually are not certain are good addresses, but the local government wants to put them in. We have gone out of the way to give them the benefit of the doubt, and so we've included them. That's in the city-style areas.

In the rural areas -- and there will be a few, therefore, in the city-style areas where this letter will simply not be deliverable. The Post Office will try to deliver it but the unit won't be there, or that address won't be attached to a unit.

In the rural areas, the Census Bureau recognizes that the postal delivery system will not reach every rural household. That's why we do our process called Update\Leave. That's why we, ourselves, starting on March 3rd, will deliver it to roughly 17 million households. We deliver it with a map spot in our geographic system. We don't deliver it with an address. That is, we have already gone across all of rural America, drawn our TIGER maps, and have map-spotted every resident. We deliver on the basis of that map spot, not on the basis of some address that's written down.

Nevertheless, it was the judgment of the Census Bureau that we should get as many of these advance letters as possible into the hands of the rural residents. And the most efficient way to do that, or the most cost-effective way to do that was to use the Postal Service as best we could. There are, after all, rural route postal address systems. There is the 911 system. And so where we could get some kind of an address, we actually had given that to the Post Office, in hopes that the advance letter will be mailed.

Our best estimate, based upon past experience is, and indeed the Post Office's own estimate, is something like 5-to-7 percent of the total address file, but a much higher percentage in the rural areas will not be mailed, will not reach the intended addressee because the address itself is simply one that we have constructed as best we can with our enumerators walking, driving the rural roads and writing it down.

Herbert Sample: Let me assume -- make an assumption here. Let's assume that most non-English-speaking residents do not live in rural areas. They live in urban areas. As you probably know, civil rights groups are a little concerned about this because this is -- one of the best ways for people who do not speak English to get a non-English-speaking, or non-English form. So would it be accurate then to say if you make that assumption that I made, that most people who need the non-English form will still get this letter, despite this problem with the address?

DR. PREWITT: Let me then back up and say the problem with the address, which we are discussing primarily today -- that is, the extra digit -- is completely unrelated to the conversation we're now having, the question you're asking. I want to speak to your question, but I just want the audience to understand, the press to understand that what we're now talking about has no relationship whatsoever to this printing error. We're now talking about the difficulty of getting an advance letter into the hands of every rural resident in the United States, which is a serious challenge.

Insofar as, that population, linguistically isolated rural households which want a questionnaire in a language other than English, if the letter does not get to them, will have to use one of our other systems. There's no other way we can do it. Our other systems include, as you know, a "Be Counted" form, which is in language. The census form itself, which will get to every one of these households because we deliver it, it has at the bottom in Spanish -- not in all five languages but in Spanish -- "If you would like this form in Spanish, please call this number." When they call that number, they can get either a "Be Counted" form, they can actually give their answer over the phone, or we can direct them to a questionnaire assistance center. All of the other systems that we have in place will then be available at that moment.

Your assumption is certainly correct. All of the demographic data the Census Bureau has available to it suggests that a very, very high percentage of the linguistically isolated households are within the city-style address universe, and so they will clearly get this letter. That's why we went as far as we could to try to include as many rural addresses as possible, also within the mail system that sends this advance letter out. But we know once we're into that system that we're stretching our capacity and the Post Office's capacity to actually identify an address to which they can mail.

MR. JOST: I just need to reminder the reporters on the phone: if you want to get in the queue you need to push 1 on your phone, and we'll go up front here and then go to a question on the phone.

Q: Dawn Holly, NBC News. What is your reaction to the Salvation Army's decision to not allow your counters into the shelters to count the people there?

DR. PREWITT: I spoke at some length this morning with Colonel Tom Lewis, who's the national secretary for the Salvation Army, and also with the legal office of the Salvation Army. Let me first read a few statements from the Salvation Army. "At the request of the Bureau of the Census, along with other religious and charitable organizations, the Salvation Army will be cooperating in the taking of the 2000 census. The Salvation Army urges its officers, employees, and beneficiaries to complete and return the census questionnaire form. A special effort should be made by the Salvation Army officers to assure the completion of the 2000 census."

This, I'm reading from the materials that they're sending out to their various outlets. "Particular attention should be given to transients, including non-resident aliens who might otherwise resist their identification to a government agency. Many social programs, including programs operated by the Salvation Army, are dependent upon an accurate census.

"With respect to the various residential programs conducted by the Salvation Army, the special places" -- which is our designation -- "will include adult rehabilitation centers, Harbor Light centers, transient lodges, residential facilities for children, other temporary housing facilities for men, women or families, in which the confidentiality is important and will be maintained by both the Census Bureau and the Salvation Army."

So essentially the Salvation Army wants to and has worked actively in a constructive partnership with the Census Bureau to account for and count all of the people in their residency programs. The issue that was reported in the press was the counting of Salvation Army dining halls and mobile food vans. Their judgment is that the possibility of a census being conducted in those areas would actually deter people from coming to get a meal, and it's their responsibility to feed as best they can.

As I said, I was talking to the Colonel this morning, and currently the plan is that we will count those persons as they're in line waiting to enter the dining halls, or waiting to get food from mobile vans. From our point of view it's not optimal, but we will still be counting them. I would prefer to have been doing it inside their facility. We are still obviously conversing with them, but as of now, the understanding is we will count all of the residences and not the people inside the dining hall. That doesn't mean we're not going to count them, but that's not where we're going to count them.

Now, let me just say one other thing, and if you want a follow-up question I'd be glad to answer it. The special enumeration process that the Census Bureau has put in place for those people without regular residencies, stresses first the shelters, which is to say where are they sheltered, whether they have a voucher to stay in a motel, for example. We first start with the shelters. Then we go to those people who do not use the shelters, where they sleep.

If we have reason to believe that a large percentage of them are in a particular park, are under a bridge, working with local advocacy groups, local city governments, we go to those areas. So we count people even if they do not have a shelter. What the count in the food system is about is a way to catch some of the people who we know are not using shelters. If they use shelters, we find them. People don't use shelters but do come and get meals, that's why we also want to count in the mobile vans and the dining halls and so forth. And we are counting in many, many of those places. Salvation Army is of course not the only place feeding the homeless, so we're counting in many, many of those places.

This is a small part of our problem. It's a serious challenge. I take it very seriously. As soon as I learned this, I was on the phone talking to the Salvation Army leadership. We concluded our phone call this morning with the understanding that we'd be back in touch. I said I was actually coming to this event, and that I would be back in touch later to continue the conversation to see if there's a way to even get closer, so that they can both do what they need to do and what we need to do.

Ms. Holly: So other soup kitchens apart from the Salvation Army have agreed to cooperate?

DR. PREWITT: We have had no one else expressing this concern at all thus far. It could happen but not yet.

MR. JOST: We'll go to the phones. We have Rick Kline of the Dallas Morning News on the phone.

Q: Rick Kline, Dallas Morning News. Yes, thank you. Two questions. First of all, how long have the advance letters been printed up, and also I was wondering if the representative of the Postal Service could talk about what the service can do to ensure delivery, what particular steps are being taken into account now that you know about this error.

DR. PREWITT: Yes, certainly. Love to have Mr. Powers join me. But let me first answer the first part of your question. The questionnaires, or this advance letter, were printed starting in, I think, late November, early December, and the print run was basically finished in the latter part of January, the third, fourth week of January. They've been printed for about a month -- a little over a month now.

I should say before I turn this over to Jack Potter -- excuse me, I said Power. It's Potter, from the Postal Service. I want to make certain that I don't leave the wrong impression, which I may have done. The Postal Service does deliver in rural areas. There are lots of people who get their mail in rural areas from the Postal Service. But the way they constructed their mailing delivery sequence, and the way in which we try to construct our addresses and our map-spotting, don't necessarily converge. That's what leads to the possibility that even as we take our system and try to meld it with their rural system, that at least some significant percentage will not be delivered that way.

That is why the Census Bureau is delivering the questionnaires itself, which is not a judgment against the Postal Service. It's simply a recognition of the fact that we have a different problem.

Let me just remind all of us once again. The census is two things. It is a count, and it's assignment of everyone to a location. The Post Office may have an address that says Rural Route 2, that is not a location. It's not a geographic location. It doesn't mean they can't deliver the mail to Rural Route 2, but number 76, but we can't take that piece of information and locate a person in a particular space of geography in this country. That's why we have to do this with a map-spot.

So we're simply talking about two systems that have somewhat different purposes. Their purpose is to get the mail out. Our purpose is to find people and assign them to a location. The census is about geography and location, not just count.

MR. POTTER: Let me continue on that point. There are some post offices in America where we do not deliver. People come to the post office and pick up their mail. So, therefore, the only location that the census would know is where the post office is. In other cases, there are clusters of mailboxes in rural America where people travel from their homes, maybe a mile, to the cluster of mailboxes to pick up their rural delivered mail. Therefore, there is no specific address.

A lot of rural America has gone to 911 addressing, so that when they call up for an emergency, they can say they live on a street, as opposed to saying my street is Rural Route whatever. So there's been a lot of effort by the rural communities around America to move to city-style addressing. Once everybody is into a city-style addressing situation, this dilemma will no longer exist.

Now, on to the bigger question. The mail that has been prepared has an 11-digit bar code on it, for the most part. That 11-digit bar code will allow the Postal Service to sort the mail, for the bulk of America, into walk sequence. The carrier will receive that mail with the rest of their bar-coded mail. The first time a carrier will see the piece of mail is when they are on the street. For example, if there's mail for 100 Main Street and it's going to John Doe, the piece of mail that's from the census that says resident, 1100 Main Street, will be sitting there right with the rest of the mail for 100 Main Street. Our carriers have been instructed to deliver that mail and ignore the first digit of the street address.

So it's not that complicated a situation. Our people have risen to the occasion many a time to handle problems. This is one that, thanks to the automation program that the Postal Service has, we're ready to handle. Our people will rise to the occasion. I'm confident of that.

MR. JOST: Would you officially introduce yourself?

MR. POTTER: I'm Jack Potter. I'm senior vice president of operations for the Postal Service.

Q: So the pressure finally goes to the mail carrier, right?

MR. POTTER: The what?

Q: The pressure to deliver will be down to the mail carrier.

MR. POTTER: There's no pressure. They'll receive the mail and they'll deliver the mail. Now what we're counting on is that the American public is made aware of this problem, and you can help us do that, so they recognize that what's important is the content of the envelope and not the address on the envelope.

MR. JOST: Any more questions for the vice president?

Q: Charles Holmes, Cox Newspapers. What was the monetary value of the contract to the printer? Were there any other printers involved in this advance letter, or is this printer doing them all?

DR. PREWITT: A single printer did this entire print run, and I believe the contract was $5.9 million.

MR. JOST: Okay, we have on the phone with us Sherry Sylvester of the San Antonio Express.

Q: Sherry Sylvester, San Antonio Express. Good morning, Dr. Prewitt. You talked about the embarrassment and the image of the Census Bureau. Last week Congresswoman Maloney held another press conference and indicated that immediately following the census, that she's still not -- she and her group are still not, satisfied that there won't need to be extensive sampling.

Do you think this letter blooper is going to assist her in making that case?

DR. PREWITT: No, I think the issue that Congresswoman Maloney was addressing had to do with whether we would get a response rate as we had hoped, as we are expecting, of not less than 61 percent. As I said just a moment ago, we are challenging the American public to help us get a response rate at 70 percent, not 61 percent. Her issue of contingency funding has to do really with the response rate.

I can see no way in which this issue that we're talking about today, in terms of the address file issue, can affect that. These are totally independent phenomena. We are confident that there are many, many things going on. Given our partnership program, our promotional efforts, the Road Van program, the Census in the Schools program, the activity of the mayors around the country and the governors, we are confident that we will not fall below our expectation. Hopefully improve upon it. There's nothing about this particular glitch that we see having any kind of consequential implications for the response rate, the mail-back response rate.

MR. JOST: If I may, the director alluded to the advertising component, and I can speak more specifically to that. We currently have in our advertising program, in-language advertising about the advance letter and how to use it to get an in-language questionnaire. There are samples in your press kit. Over the weekend we have supplemented that advertising program and are producing additional radio and print advertising that seeks to address any of the confusion that might come to the public with respect to this mailing, and whether or not it will be delivered to the intended recipient.

You can see behind me one of the posters that we've already developed to encourage people to open the mail and to read the instructions and respond to the instructions if it affects them.

Q: Genaro Armas, AP. The quality control procedures for the letter, are they the same ones used for the questionnaires themselves, and do you think any changes need to be made?

DR. PREWITT: No. As I said, we spent considerable effort in the last several days reviewing both the quality control procedure, but also re-examining the print runs that have already taken place with respect to the questionnaires, the forms that have all been printed and are going into the mail stream even as we speak. The update/leave, are obviously already being delivered to our local offices so that we ourselves do the delivery in the update/leave.

We have re-subjected all those print runs, all the questionnaire forms to all of our checks, plus some additional checks, and we have no reason to believe that there will be any carryover whatsoever. We can't even imagine what the carryover would be between this particular printing glitch and the census forms themselves.

We run things through the processes more than once. We have different kinds of people re-running them. We have different kinds of visual checks. So all the kinds of things that we have already done with those print runs, we've re-done plus additional things, and we're finding absolutely no problem.

Q: Is there someone here from GPO who could come up and address what they're doing to look into this?

MR. JOST: Yes. We have Andy Sherman, the director of public affairs and congressional affairs.

Q: I guess I'd like to ask what you're doing to investigate this, whether you could retrieve any money from the printer for this. And then a little bit about your news release mentions that this printer has done other jobs that have gone well for the government. Have they been of this complexity and magnitude, and could you cite a couple of examples?

MR. SHERMAN: I'm Andy Sherman. I'm the director of public affairs for the Government Printing Office. With me is Jim Bradley, who is the director of customer service for the Government Printing Office. You have before you in your kits a GPO statement on the letter. I just want to say at the outset that Freedom Graphic Systems, located in Milton, Wisconsin, is a very reputable, experienced commercial printer who has done other government work before for the Office of Personnel Management, for the Social Security Administration and, we think, some IRS work, as well.

He won this contract on a competitive basis on significant competition last summer. We are working with him now. We got back in contact with him immediately last week, as soon as we heard from the Census Bureau that there was a problem with the addresses. We're working with him to find out exactly what happened.

When he was awarded the contract, we conducted what was called a pre-award test of the addressing of the letters, based on the databases that we were provided -- or that were provided to him for these letters. The pre-award test ran fine. It ran successfully as to all of the conditions that need to be met from a printing standpoint, including positioning of the image, ink density, spacing and so forth, and as to content, as well. So the tests ran fine.

What we are trying to determine with him now is where exactly, or how exactly this error happened.

The contract with Freedom Graphic Systems is between the Government Printing Office and that company. We have, as agents for the U.S. government, certain rights and responsibilities under that contract, but so does he. And so what we are trying to do is determine with him where exactly the problem lay. Then we will work with him from then on to determine what can or should be done about it.

Q: Are they cooperating?

MR. SHERMAN: Yes, they are cooperating fully.

Q: [Inaudible.]

MR. SHERMAN: We have certain options that are available to us under government contract law and government procurement law, depending on what we find. But as Dr. Prewitt suggested, it's premature at this time to assign any blame until we know exactly what the facts are.

Q: [Inaudible.]

MR. SHERMAN: Printing is not a perfect process, as all those who are involved with newspapers know. It is unusual for a problem of this scope to materialize, and we want to emphasize, as Dr. Prewitt has, that as soon as we found the problem with this addressing issue, we went to all of our contractors who were handling all our census materials, particularly the data collection instruments, and reaffirmed with them, rechecked with them that there is not a problem of this kind with any of those other documents. And we have no indication at this time that any such problem exists anywhere other than with this particular job.

That's why we said in the statement that this is an isolated occurrence as far as we're concerned, with the knowledge that we have at this point in time.

DR. PREWITT: I certainly did not at all mean to do anything other than treat this issue very seriously. We treat it very seriously. However, as Mr. Sherman says, print runs are print runs. I have before me an article from the New York Times that says the Census Bureau begins count with 120 million wrong addresses. I have an article on the New York Times Web page that says, Census addresses to go to 12 million wrong addresses. I like the latter one better than the former one.

MR. SHERMAN: I just want to add, while we have the privilege of attending this press conference -- this is probably the 12th or 13th decennial census that the Government Printing Office has supported the Census Bureau on. The census work constitutes the single largest group of work in the Government Printing Office at this point in time. We treat this matter extremely seriously. We know how important the accuracy of the materials is to the decennial census -- we reacted, we feel, promptly when we learned from the Census Bureau about what had happened. We are taking every step that we can conceivably take to ensure that all the materials that are coming out from here on will not have this problem.

Q: Was the contract $5.9 million? There was another figure floating around.

MR. SHERMAN: The Washington Post reported $5.6 million, we saw. The cost to the government will be $5.9 million.

DR. PREWITT: The Washington Post reported correctly what they were told.

Q: Was it Milton, Wisconsin?

MR. SHERMAN: Milton, Wisconsin, yes. The contract between the Government Printing Office and Freedom Graphics is a matter of public record. We'd be happy to provide that information to anybody who's interested.

Q: Ann Collins, New York Times. How do you know that the Census Bureau systems didn't somehow spit out faulty numbers?

MR. SHERMAN: Well, as I say, we ran a pre-award test which was used to determine the accuracy and the quality of the information that was provided to the contractor and it ran absolutely successfully.

Q: Is the presumption that the mistake occurred in the software, and it's a matter of finding out how that got in?

MR. SHERMAN: We're not convinced that it was in the software at this point in time. I don't think we can make that statement. We need to finalize our investigation before we come out with any assignment of responsibility.

Q: Is there any indication that it was tampered with?

MR. SHERMAN: There's no indication whatsoever that it was tampered with. No indication whatsoever.

MR. JOST: Thank you, John.

MR. SHERMAN: If any of you have any further questions, we'll be around for a little while to answer anything else.

MR. JOST: We'll go back to our phones. We have Ellyn Ferguson with Gannett News Service.

Q: Ellyn Ferguson, Gannett News Service. Hi. I have two questions. One, how was the mistake discovered, and two, you spoke earlier about public perception. What does a mistake like this do to public perception as you're starting to gear up for the main part of the census?

DR. PREWITT: It was discovered, and I can't give you the exact location, but it was at a post office in New England -- I think in Maine, but we're not certain. It was in New England. They simply gave us a call, said, look, or gave our local office a call, saying they'd just received the advance letters, that they didn't recognize all the addresses and did we have any explanation for that. That triggered the investigation that started Wednesday morning. We got this call late Tuesday evening, and then we went into high gear, of course, to try to establish exactly what could be known.

I do want to reiterate what was just said by Mr. Sherman, that the GPO was quick on this, was very cooperative with us, as was, of course, the Post Office. So this has been the kind of thing the government ought to do. When a mistake happens of this sort, you want to first make certain that it doesn't have any operational implications or consequences for the task before us. The task before us is to make sure that we have a good census. And only a second order issue, though obviously still important, is to determine just exactly how it happened. Our first focus was on would this affect the census and, if so, what would we have to do to compensate for it.

We discovered fairly quickly in our conversation with the Post Office that it will not affect the census, and that's the major message that I want to stress today.

Your second question was?

Ferguson: You mentioned earlier the public perception.

DR. PREWITT: Of course. Certainly. Well, I don't want to sit here and spin theories about how the public will or will not respond to this. It partly depends upon how they learn about it, how much press coverage there is, the quality of the press coverage. We'll live with whatever the press does, of course, but it will partly depend on how it is argued to the American people.

I take very seriously the Census Bureau's obligation to maintain its credibility with the American public. We are the premier collector of quality statistics for this government. We're recognized in that capacity around the world, and everything that we can do to demonstrate that we are a professional, quality operation that delivers good statistics to this country is critical. Therefore, when a mistake like this comes along, it is a public image mistake. But I take it very seriously because we really do want to communicate the truth of the matter, which is that the Census Bureau is an efficient, effective organization. It does get good numbers. It will report those numbers to the country. We want the census to be a success.

That's why we have gone into high gear over the last 48, 72 hours with the kinds of programs that I've already alluded to and that Steve emphasized, this poster stressing the importance of this letter. Major work this morning with our partners. We have a major partner meeting going on right now in Baltimore. We're using this as an opportunity to get actually more attention to the advance letter than it might otherwise have gotten.

So I can't really say exactly how the public will respond to this. I hope they simply recognize that these kinds of glitches do occur, and I hope they are reassured and relieved that it has no consequences for how well we will count the American people. Indeed, what we hope is that more people will pay attention to the advance letter than otherwise might have. As I said in my opening comments, our expectation is that the advance letter does increase mail-back response rate, or else we wouldn't be doing it. We're doing it for a purpose: to increase mail-back response rate, and also, of course, to provide the opportunity to get the language form.

MR. JOST: We have time for just a couple more, if there are any more questions.

Genaro Armas: Going back to the Salvation Army. How much of a hindrance do you see that, their policy, to be as far as to getting an accurate count of the homeless population?

DR. PREWITT: Well, let me just repeat. The basic count of that population depends upon our going to the shelters and finding the people in the shelters. That is, people both sleep and eat. Everyone in the country does, whether you're homeless or not. Everyone both sleeps and eats. So our primary count is based on where people sleep, that's true for how we're going to count you. We're not into where you eat. We're interested in where you sleep because this is a residency-based count, of course. It is a geographic location.

Therefore, our fundamental strategy is to go where people sleep or get shelter. We know that will not catch all people because not everyone uses shelters, so we also have in place a process to go to where people who do not use shelters nevertheless sleep -- in parks, certain locations. And we will attempt to count everyone there, as well, in this operation.

Like so many things we do in the census, we actually try to be as careful and exhaustive as we possibly can. So counting people in the food line or in a dining hall is just an extra step to what we've already done. We asked the people when they're in the food line, or when they're getting a meal, "did you use a shelter last night?" If they did, that's where we count them. That is, we don't double count them. If they were in a shelter, we presume that our shelter enumeration caught them. So we can't double count them by not counting them where they eat.

So this is only for that residual part of the population which does not use a shelter but does come in for a warm meal and then goes back to the streets. We may catch them on the streets, but this is just an extra.

Now, if none of the places who provide food to the homeless are willing to cooperate with us, that would obviously hurt the census. We wish it would not happen. In fact that it's only one provider, and that provider is a particular provider where people do stand in line to get in. We will count them in line, it's a fairly small part of the population we hope will be affected. Nevertheless, we take everything about the census seriously that can possibly lead to any undercoverage. That's why we continually add operations to improve coverage. And as I said before, I will continue to talk to the Salvation Army, hoping that they will agree, as so many other places have, to allow us actually to use the spaces themselves. That may or may not work. We'll see.

MR. JOST: Anything else? Any further remarks?

DR. PREWITT: Yes. I would then, if I could, just stress to all what is really critical at this point in the census -- we start update/leave in just a few days, the end of this week. The advance letter is intended to tell you, open this envelope, read it, understand that the census is important. This is the message. Open this letter when you receive it. We believe that the American people will do that and that will help us move along as we move into the census process.

Thank you for your time.

[END OF PRESS BRIEFING.]