A Way Into the 9/11 Air Records

If you were there

The World Trade Center Health Program provides monitoring and treatment at no cost and was still enrolling as of September 2026. You may be eligible if you were a responder. You may also be eligible if you lived, worked, went to school or were present south of Houston Street, or in parts of Brooklyn, on 11 September 2001 or in the months after. Survivors are now the larger source of new applications.

The September 11th Victim Compensation Fund is separate, and covers economic loss and physical harm.

Neither depends on finding your building in these records. A document is not required, and its absence proves nothing.

The city has also said it will place staff at two agencies to help people use these records to access the federal health and compensation programmes.

What was measured

Coverage has focused on what officials said against what they knew, and on asbestos, though benzene and the Staten Island landfill have been reported too. The measurements themselves have had less attention, and asbestos was a minority of the work. The city’s laboratories also tested for dioxin, lead, PAHs, silica, fibrous glass and mineral fibres, in air, wipes and vacuumed dust. Solvents and metals fill 29 of the 68 boxes.

Where the last column says complete, the figure is the highest that exists in that record. Everywhere else it is the highest in what I happened to read, and a fuller search would very likely raise it.

Every reading, filterable by where it was taken and shown against its own limit, is on the archive page.

Whether the air was safe depended on which yardstick you picked

There was no limit for asbestos in outdoor air in 2001, and there still isn't. The city said so itself, in a briefing note explaining that the figure it used — 0.01 fibres per cubic centimetre — was written for indoor spaces and used because nothing else existed. The same is true across the list: the thresholds applied came from workplaces, from indoor clearance rules, or from guidelines written for lifelong exposure.

Of the 130 street samples the city counted between 15 and 23 September 2001, 53 were over the indoor clearance figure and none were over the limit for a worker’s eight-hour day. The same air, the same samples, and two defensible answers about forty times apart. Which threshold an official reached for decided what could truthfully be said that week.

Reading the chart: the groups cannot be set against each other. A workplace limit is set close to where harm is expected, for healthy adults in protective equipment over eight hours. A public-health guideline is set for everyone breathing continuously, with safety margins often a hundred to a thousand times wider. Thirty times a guideline may matter less than five times a workplace limit, and twice a limit is not twice the damage.

What they found

The nine

The city’s own summary reduces the street monitoring to one figure: nine samples out of 4,564 exceeded the asbestos benchmark, or 0.2 per cent. Which nine, and where, is set out on a single page of the city’s own monitoring file. The nine came from five places. Murray and Church accounts for four of them, on four days inside a week; Warren and North End Avenue for two. The city’s own site numbers — 8, 11, 12, 19 and 32 — are on the same page, so a disputed corner can be checked by number rather than by name.

All nine fall in ten days of September. Murray and Church accounts for four of them, on four separate days. None is more than about twice the limit.

The same page gives the denominator: 3,540 samples from 37 sites around Lower Manhattan, 616 outside four schools, 408 in the outer boroughs. The same page also records 19 overload sites. An overload means so much material collected on the filter that the laboratory could not count it. No reading could be taken from these, so they count neither as exceedances nor as clean results, and the 0.2 per cent leaves them out.

The schools were watched daily — starting three weeks late

Daily sampling outside four downtown schools began on 4 October 2001. By then the towers had been down for three weeks and the worst of the plume had passed. The dust cloud people breathed on the day was long gone, and no monitor had been running in it. The four buildings are several blocks from the site. And the substance measured was asbestos, which the health record suggests was not the worst of what was in the air. Nine in ten of the readings came back below what the laboratory could quantify.

Within those limits, the series is the most complete run of measurements in the release: every day from 4 October 2001 to 30 June 2002, of which I extracted 1,079 of 1,080 cells. One exceeded a limit — I.S. 89 on Warren Street, 7 October 2001, at 0.018 f/cc — and the same cell carries an electron-microscope count of 40 structures per square millimetre, against a benchmark of 70. The city published its own count for those schools as of 6 March 2002. Cutting my series at the same date gives 616 readings, and so does theirs.

This says nothing about whether anyone was harmed. Children who attended these schools are eligible for the health programme, and survivors — residents, workers and students — are now its fastest-growing group. Clean readings outside a building three weeks afterwards do not describe what was breathed on 11 September, or indoors, or on the way to school.

Solvents depended entirely on how you measured

A few minutes of air taken at the pile gave benzene at 49 parts per million, against a workplace limit of one. Sampled continuously over twenty-four hours a few streets away, the same compound came back at five thousandths of a part per million. The two are about ten thousand times apart because they measure different things: minutes of air at the pile against a whole day’s average a few streets away. Neither is wrong, and they cannot be compared.

Staten Island, where the debris went

The wreckage was barged to the Fresh Kills landfill and unloaded there for months. The federal database holds 5,207 measurements across the landfill sites, and 50 of them are above the asbestos benchmark, against 22 of 8,870 in lower Manhattan in the same evaluation. Thirty-six fall in October and November 2001, when most of the unloading was done, and the highest is 275.56 structures per square millimetre on 5 November. That is nearly four times the benchmark, and higher than any reading taken in Manhattan.

The federal evaluation judged that worker exposure at the landfill was minimal and that no occupational limit was breached. Both things are in the same document: the highest asbestos readings in the federal database, and a conclusion that the people working there were not harmed by them.

EPA sampled at the landfill from 19 September into mid-October and reported the results to the city’s health department, so they sit in the same monitoring file as the Manhattan readings. Two of eighteen samples on 19 September were above 0.01 fibres per cubic centimetre, at a maximum of 0.029; everything reported from 22 September on was below the detection limit or below the 70 structures per square millimetre standard. The city’s own network, 3,540 street sites and 616 school samples, does not reach Staten Island.

What people were breathing through

The respirators

Measurements describe the air. What reached people’s lungs also depended on whether they were wearing anything. A federal inspection report from October 2001 found that at Ground Zero respiratory protection was rare except among heavy equipment operators, that workers were seen in the smoke plume without hard hats, eye protection or respirators, that most did not use the wash stations on leaving, and that no safety and health programme appeared to be operating at the site. Use improved from around 22 September.

At the Staten Island landfill it was different. Workers there were told that not wearing a respirator would bring disciplinary action, and federal investigators who were present confirmed the contrast between the two sites. So the site with the highest asbestos readings in the federal database also had the strictest protection, while Ground Zero, with lower readings, had almost none.

The consequences were being recorded within months. A study of firefighters found that 8 per cent of those present at the collapse developed a cough severe enough to need at least four weeks of medical leave, against 3 per cent of those arriving within two days and 1 per cent within a week.

The dust, which no monitor recorded

Every number on this page came from an instrument. The largest exposure has no number, because no instrument was running in it. People stood in the cloud for hours on 11 September and nothing was running in it. The federal evaluation says it cannot be stated with certainty what followed for those who were engulfed.

Keep that in mind for everything below. The illnesses that dominate the health programme — sinus disease, reflux, asthma — are ones physicians have attributed to inhaling and swallowing caustic dust rather than to asbestos. The record says least about the exposure that best fits those illnesses.

What the dust actually was

Most of what is measured here is asbestos and solvents, because those are what the monitoring programme was built around. The dust may matter more. Tests by the United States Geological Survey found it strongly alkaline — in places as caustic as drain cleaner, capable of severe irritation and burns. That is a plausible mechanism for the illnesses that dominate the health programme: sinus disease and asthma are injuries of the airway, where large caustic particles land, and reflux is an injury of the upper digestive tract, where swallowed dust goes.

This was disputed. Separate tests at New York University agreed the larger particles were highly alkaline but found the smaller ones — those able to reach deep into the lungs — were not. The Geological Survey’s tests had not separated the dust by particle size. An EPA chemist challenged the university’s findings; the university and the agency stood by them.

No pH figure appears in the readings on this page. It is not among the substances listed in the city’s own summary of what it sampled for, and I have not found a measurement in any document I have read. I have not searched the whole archive for one.

What was said, while the results were pending

Every statement below is dated, public at the time, and in the city’s files. Read against the laboratory dates, most of the autumn’s reassurance was issued before the results for those days existed.

In one laboratory, results came back weeks after the air

One section of the city’s laboratory tested for solvents and combustion products — benzene, toluene and the like — using a technique called gas chromatography–mass spectrometry, which separates a sample into its compounds and identifies each by weight. Its folders record when a sample was collected and when it was read; 260 of 272 do, so this is nearly a complete record for that section. The median gap was 19 days in October 2001 and 27 in November, fell to 14 in December, returned to 27 in January, and reached four days by May 2002.

This covers that one section. The metals laboratory wrote a single date on its folders, so its turnaround cannot be worked out, and may have been different.

The gap was not constant. The pesticides section, the only one with September batches carrying both dates, turned seven of them round in six to eleven days. The solvents section then ran at 19 and 27 days through October and November as the volume built, and came back to three days by the following May. A laboratory slowing as work piles up says nothing about what officials were telling people.

The dates do limit what could have been known. The towers came down on 11 September, and at any turnaround in this file no post-collapse solvent result existed when the first reassurances were given a week later. Later statements that autumn used air sampled weeks earlier. The records do not say whether anyone noticed.

What X stands for is not settled. A sample-log legend inside NYC-WTC_000148192, page 421, gives the codes as Collected and Analyzed, and the chart reads them that way. But that legend belongs to the pilot cleaning log, not to the laboratory boxes, and the folder names point the other way. Sixteen spell the step out — 508 9/20/01 Ext. 9/28-9/29 Analysis — a bare collection date, then extraction, then analysis. The 333 that use X carry no bare date before it: GCMS Scan X 10/10/01 An 11/1/01. In 624 folders, X and a spelled-out Ext never appear together. That fits X meaning extraction, which happens after a sample reaches the laboratory.

Both readings point the same way. If X is collection, the gap below is the wait from street to result. If it is extraction, collection came earlier and the wait was longer. The chart shows the shorter of the two. The folder names are in the data file.

What the cleanup was arguing about, inside the city’s files

The indoor cleanup box holds about 790 documents, mostly short: working-group meeting notes, scopes of work, the funding matrix sent to the federal emergency agency, and a folder of memoranda from Cate Jenkins, an EPA chemist who spent 2002 disputing how the cleanup was designed. The note about the word “risk” came from this box. For the indoor programme, it is the nearest thing the release holds to a record of how decisions were reached.

One of Jenkins’ arguments bears on the readings above: that the threshold used for dust — one per cent asbestos — was taken from demolition rules and measures a proportion, when what matters on a floor or a carpet is how much sits on a given area. The air thresholds were borrowed in the same way. These documents do not show whether she was right, only that the city was circulating and filing what she wrote.

“We would not use the word ‘risk’”

A one-page working note dated 16 May 2002, headed Clean-up Initiative, records four conversations about the federal programme to clean apartments in Lower Manhattan. The last item concerns letters the city was to send supporting the funding request. Its final sentence: the writer stressed that the word “risk” would not be used in any of the letters, as it would defeat the purpose of creating public confidence.

An earlier item in the same note is less quoted and harder to read charitably. The City Law Department wanted testing done before any cleaning. The writer objected that if testing came first, the federal emergency agency might refuse to pay for a large cleanup if the testing showed everything was fine. Testing first was a risk to the funding.

The note exists in two identical copies in the same box, so the wording is not a transcription artefact. It is a staffer’s working note, not a policy instruction, and it records what one person stressed to another.

Who was left out

The carpets

On 9 June 2002 Jenkins circulated a memorandum headed “Testing Carpet, the Asbestos Reservoir”. Its argument: carpets and fabrics hold asbestos and keep releasing it, the agency’s own studies show carpet cannot be decontaminated by HEPA vacuuming even with water or steam extraction, and a better test existed — the agency’s own ultrasonication method, needing no consultant and costing only a laboratory fee. She asked why it was not being offered to residents during the free testing and cleanup below Canal Street.

The next morning the memorandum was forwarded inside the federal agency to the officials running that cleanup, copied to a city address, and reached the city’s environmental department the same day. The city’s copy is in its files, stamped with the date it arrived.

The cleanup went ahead using HEPA vacuuming. I cannot settle whether the criticism was right. The city received it on 10 June 2002.

Chinatown and the line at Canal Street

The federal cleanup programme covered residences south of Canal Street in Manhattan. Everything north and east of that line was outside it. In a submission kept in the city’s files, Rep. Jerrold Nadler argued that this left out places that had been under the dust cloud — the Lower East Side, Chinatown, and parts of Brooklyn — and cited the federal inspector general’s view that the cleanup area should have been set by sampling outward from Ground Zero in widening circles until the edge of the contamination was found.

The city’s own sampling records are consistent with the complaint. Three monitoring locations in the daily tables lie east of Broadway, and none are in Chinatown or on the Lower East Side.

What happened there was measured later, by clinicians rather than by monitors. A study reported to the City Council looked at children treated at a community health centre in Chinatown, about a mile and a half from the site. In the year before the attacks 306 children with asthma made 1,044 visits; in the year after, 510 children made 1,554. The researchers found a similar rise among adults.

Language was part of it. A council submission argued that an announcement in English and Spanish about the asbestos in the dust could have made a difference to the day labourers hired to clean lower Manhattan offices without training or equipment. By May 2002 the federal cleanup hotline had representatives speaking English, Spanish and Chinese, and the web form was in English with Spanish promised.

No street monitor in Chinatown, and three east of Broadway

The same submissions made a second complaint. A council member argued in 2002 that the Lower East Side and Brooklyn had been left out of the testing altogether. The daily tables show where the monitors were, so the complaint can be checked against them. I searched all 428 pages of the city’s monitoring file for the street names of Chinatown and the Lower East Side — Mott, Bowery, Mulberry, Elizabeth, Essex, Pike, Catherine, Madison, Monroe, Cherry, Henry, Clinton, East Broadway. None appears anywhere in it.

The daily tables name 311 sampling locations across 1,855 rows, though many are spellings of the same corner: the ambient network merges to 45 places, and the city describes it as 37 sites. Three of those places lie east of Broadway — the corners of Centre and Chambers and of Spruce and Gold, and one building, 100 Gold Street, sampled indoors at the lobby and the ninth floor. They account for 29 of those 1,855 rows. My parse is incomplete — it covers 14 September to 10 October 2001 cleanly, and the later pages, which run to March 2002, are scanned too poorly for the same method. Spot-checking those later pages recovers 31 more location names, all of them west. How many places that is out of depends on how you merge the names. What does not: no outdoor street monitor stood in Chinatown or on the Lower East Side, under any way of counting.

Of the 721 placed building addresses, 337 lie east of Broadway and hold 1,863 documents — Water Street, Front Street, Broad Street, Wall Street, the Seaport. Eight sit on Chinatown and Lower East Side streets, six of them on Madison Street. The city held files on buildings there, but no street monitor. The complaint was about monitoring.

Two indoor samples in the file do sit east of Broadway — City Hall on 16 September, at 0.022 f/cc on the second floor and 0.011 in the Blue Room, and a departmental pumping station at 184 Avenue D on 14 September. Both are inside city buildings rather than on the street, and the City Hall reading is twice the reoccupancy threshold. A second set of 21 monitoring locations, from laboratory certificates in the email boxes, could not be placed. Their names put them all in Battery Park City and the streets west of Church, which is a guess from the name, not a measured position.

There is nothing on Canal, Mott, Mulberry, Bowery, Elizabeth, Allen, Essex, Pike, East Broadway, Catherine, Madison, Monroe, Cherry, Division, Henry or Clinton. Not one row in Chinatown or the Lower East Side.

Both outdoor eastern sites read above the asbestos benchmark on 12 September, at 123.73 for Centre and Chambers and 157.48 for Spruce and Gold. Neither appears on the list of nine above. Nor do 131.23 and 121.69 at Fulton between Church and Broadway, 102.36 at Washington and Park Place, or 80.00 at West and Albany, all west of Broadway and all inside the period the city’s summary covers.

The 12 September samples carry an earlier series number than any of the nine, which begin on 15 September. But the nine themselves are drawn from several different series, so the numbering is not what decided which readings were counted. I cannot establish from the pages I looked at what separates them. The same two eastern figures are the ones Jenkins cited in 2004, in an open allegation.

What I don’t know

What the email boxes hold

Four boxes are departmental email: about 3,300 index entries, mostly the sent and received mail of two DEP staff between 2001 and 2003. This is the closest the release comes to a record of what the city knew and when. Enumerating folder by folder reached 1,370 of them.

Twelve of the fifteen folders are labelled as correspondence; one is labelled meetings and holds minutes and proposals. I read 34 documents, 28 of them from that folder, which is where the decisions turned out to be. The other thirteen folders I have barely opened.

A task force proposal lists among its activities the digitising of results from more than 100,000 samples taken by contractors in private buildings, with the stated purpose of sharing them to reassure the public — followed immediately by a note to check with departmental lawyers first. The same document proposes that the Deputy Mayor for Economic Development lead what it calls a confidence-building initiative, and suggests a launch event with the Mayor to draw attention to it.

Behind that proposal is an inter-agency structure. One set of minutes lists a legal working group coordinated from the Office of Emergency Management, bringing together the general counsels of that office, the Law Department, environmental protection, health, buildings and housing. Its listed subjects are responding to complaints and testimony, press announcements, timing, a single place for information, and public participation.

The chemist who alleged that readings had been removed appears in these minutes as a press problem. Minutes of a working group record officials, including one from the state environmental agency, objecting that she had been presented as speaking for the federal agency, and asking why it had not moved to refute the report.

I found no one discussing whether to publish a particular measurement, which is the question these boxes were most likely to answer.

Not all of it is correspondence. A 27-page document filed under 2002 email is a run of laboratory certificates: 287 air readings at 21 named corners between 6 January and 1 April 2002, with flow rates, volumes, fibre counts and detection limits. One is at or above the reoccupancy threshold — 0.044 f/cc at Vesey and River Terrace on 31 January 2002, four and a half times the threshold and the highest outdoor reading anywhere in this project’s data.

That reading is a single sample from one contractor’s certificate. It is not among the city’s nine, and does not contradict them: the nine are electron-microscope counts over 70 structures per square millimetre, this is a fibre count over 0.01 f/cc. Different method, different benchmark. It shows that measurements above the threshold were still being recorded in Lower Manhattan in 2002, and that some of them sit in boxes catalogued as email.

Most of the archive is unread

I have opened 118 documents of roughly 24,450, and which 118 was not a systematic choice — see how far to trust this. Four of them supplied almost every reading on this page: the school table, the daily monitoring reports, one federal evaluation, and one building’s close-out file. Two more datasets — laboratory turnaround and the address index — were built from folder labels without opening anything. Counting how often words appear across the whole index gives a rough sense of the rest:

Each square is one box. Nearly half are laboratory paper and two hold only timesheets. Of the documents I fetched, I read a minority closely; the rest I only searched. Six such documents have no box recorded in the portal and are not on the map. Hover, tap or focus a square for its number, what it holds, and how many documents came out of it. More on what is in the boxes

A word appearing is not a document being about it, and shares overlap. Even so: lead appears in a fifth of the archive and I hold ten readings of it, all from one table. Calibration records are almost as common as asbestos, which says how much of this is laboratory paperwork rather than findings.

Reported elsewhere and not yet read here: lead contamination at the Department of Transportation building — "Department of Transportation" lead returns 123 index entries, none opened. The World Trade Center 7 collection is 6,050 entries of its own, of which I have read none; the few mentions I have seen elsewhere include hazmat teams taking wipe samples from transformers buried in its rubble, a known source of PCBs. The New York Times has reported on an audit from November 2001, prepared by a private firm and given to the federal environmental agency, finding benzene still spiking near the towers’ footprint and asbestos at the Staten Island landfill rising after the attacks. I searched for it — "audit" November 2001 EPA benzene and four variants — and could not identify it. It may be there under a title I did not guess. Specifically thin: the indoor cleanup box, 790 documents, of which I have read six and catalogued the folders; four boxes of departmental email, of which I have read 34 documents, all but six from one folder; and the great majority of the building files. The city says this first release is the beginning of millions of pages: a large collection sent years ago to the insurer defending the city against 9/11 litigation was recovered in 2026, and further material is under review. A day-by-day series of street readings has been extracted but does not yet reconcile with the city's own sample count, so none of it appears here.

The city’s own summary gives its street monitoring network as 3,540 samples between 11 September 2001 and 6 March 2002. I hold 1,481, at 45 locations, and cannot account for the difference. It is not a missing part of the file, because I extracted all 428 pages and the pages my parser reads poorly hold the same western locations. Nor is it the date range: the sampling was overwhelmingly in September, and September is what I have. Most likely the city’s figure counts a wider programme than this one document records, but I have not shown that, so treat my sample counts as a minimum.

Eleven monitoring locations could not be placed, so they are missing from the maps though present in the data; most are Battery Park City parks and waterfront. The perimeter sampling tables are extracted but unpublished: only 91 of 594 rows carry a value, and some are mislabelled by method, which the claim audit explains. Thirteen of the fifteen email folders remain barely read.

I read all 428 pages as text and searched them for the street names of Chinatown and the Lower East Side — Mott, Bowery, Mulberry, Elizabeth, Baxter, Essex, Allen, Pike, Catherine, Madison, Monroe, Cherry, Henry, Clinton, Division, East Broadway. Then for the landmarks a site might be named after instead: Confucius Plaza, Seward Park, Chatham Square, Columbus Park, the schools east of Broadway. That second search found City Hall, which had been in the indoor readings all along and which I had missed.

I searched the whole portal too, for the same names paired with monitoring terms. Those queries return hits, but the portal does not honour quoted phrases — a three-word phrase returns thousands of documents matching any one word — and the two most promising results contained none of the terms when I opened them. The portal can neither confirm eastern monitoring elsewhere nor rule it out, so the claim here is about the city’s daily monitoring file only.

And an open allegation

Jenkins alleged in 2004 that high asbestos readings east of Broadway were deleted before the city published its data, citing two figures at Centre & Chambers and Spruce & Gold. Both are in the city’s internal monitoring file at the values and locations she gave, so the underlying data says what she said it said. Her actual claim — that they were removed from the version published in early 2002 — cannot be tested here, because that published version is not in this archive. The city’s summary also leaves out four readings over the benchmark west of Broadway, so the omission she describes is not confined to the east.

Now look up your own building

The city opened files on buildings across Lower Manhattan, mostly contractor records: inspections, abatement paperwork, and air monitoring worn by the people doing the cleaning. I have matched 824 addresses to them, merged from 1,020 folder entries and covering 4,564 documents. A file existing does not mean a building was contaminated, and none does not mean it was clean. About a third of the building documents are not yet matched to an address, so a blank result may mean the file exists and I have not linked it. The first document at an address is often a one-page portal cover sheet.

How far to trust this

It is a gist. What is here reflects where I happened to look: a document opened for one reason that turned out to hold something else, a folder read because its label sounded promising, searches phrased in the language of alarm that return documents where someone was alarmed. The coverage numbers above say how thin the reading is. They do not say which way it leans.

It is also one person’s reading of scanned paper, most of it pulled out by machine from faxes of typed tables. That reading can be wrong, and has been: a parser ate a decimal point and turned 0.018 into 18, a table was read in the wrong orientation, a count of missing days went in at 57 when it was 42. Each was caught, but only because I went looking.

Most of what I pulled off these scans cannot be checked, because the document exists in one copy. The school tables are the exception. The release contains three separate filings of them, so I could check my reading against a second scan of the same page. Two of those filings cover part of the run, 446 of the 1,079 readings, and of those 431 matched: about 96.6 per cent. All fifteen disagreements are the same mistake in the same direction — a faint “less than” sign read as a measurement, turning “below 13.33” into “13.33” — and all sit at the detection floor, so none changes a finding. The inspector general’s report and the May 2002 memo also exist in duplicate here, and both carry their quoted passages identically.

That is all the verification there is. Every other figure on this page rests on the assumption that the same method failed at about the same rate elsewhere.

Every figure from a document links to its portal page, except one held back for privacy; it appears in the data by number. I reported the problem to the city’s Law Department before publishing. The health-programme numbers link to the CDC; the turnaround figures and the address index were derived from folder labels rather than pages, and are labelled as such. The extracted data, the code that produced it, and a dated list of every error found so far are all downloadable below. How the archive is organised, how the numbers were read, and what is new here versus long public: Inside the archive.