Frequently Asked Questions

A: We've tested the anthology on the following browsers and have noted the following characteristics.

Microsoft Internet Explorer Mac (obsolete as of January 31, 2006): The Keywords window formats to two columns, but is functional.

Camino 1.02
FireFox 1.5.0.4
NetScape 7.2: OK, but some tables show their borders. Looks kind of cool though.

A: The early (pre-1995) NTSB reports are all scanned from documents at 225 or 300 dpi. Then text is extracted by optical character recognition, then the original document is downsampled to 150 dpi. Each page is thus a picture, along with information needed by the PDF file to maintain a text representation of the document. This makes these documents much larger than more modern reports, which were created on computers and simply converted to PDF.

Note that since machine translation was used, sometimes there will be discrepancies between "what you see" and "what you search for."

In general, with many many caveats, it should be possible to print any report in this anthology on a modern printer with a performance of one page per minute or better. The reports that will print the slowest will be the old (big) ones, each page of which are large pictures. Modern reports will print much faster.

Printers which use the host computer to "render" the image, then print the image line by line, should do well. This includes most inkjet printers.

Modern printers which have a lot of onboard memory and their own processor should also do well. This includes most black and white laser printers. "Color lasers" require more processing power (thus more time), so if you print the images in color, they may take more time to print.

1990s-era laser printers, which tended to have very small amount of onboard memory and very slow processors, will work, but it may take a while to print each page, particularly the older big reports. Like half an hour a page.

1990s-era "color lasers" are also likely to be very slow. In one test, a modern sub-$80 inkjet greatly outperformed a company's 1998 flagship color printer by huge margins, both in terms of quality and speed.

The last variable is how you're connecting with your printer. Again, looking at the large files, since the rendered graphics have to be sent to the printer, the faster the connection, the better. Hardwired Ethernet (or equivalent) the fastest, followed by wireless, USB. And technologies nobody really use for printing anymore, like SCSI, serial, and LocalTalk.

Acrobat should properly scale and rotate pages when printing. If you find that you're printing "postage stamps" please check your "Page Setup" settings and the output settings in Acrobat or your PDF viewer. "Scale to Page" should be selected.

A: Yes. Acrobat has a powerful navigation feature, "Bookmarks." This is a collapsible table of contents, with links to the relevant sections. Few published accident reports take full advantage of this feature. This feature was added to every report, in a consistent manner.

In one case, the 1978 PSA midair, we were unable to find the copyright holder of the photographs included in the report, so fuzzed them out. These images are widely available on the Internet.

In the case of Jeppesen charts, we were also obligated to blank them out. See the Jeppesen section below for work-arounds.

Where there's introduced content, there is no mistaking it, as any editorial notes are in color, usually on a bright yellow background.

A: We were unable to obtain clearance from Jeppesen-Sanderson to reprint their charts. Without their permission, we were unable to reproduce these materials, which are clearly labelled as copyrighted by them in the original materials. We thus blanked out the images where they occurred. This is unfortunate, as it diminishes the historical reference value of these particular documents. This mainly affects about 40 reports out of the 1125 in the collection.

A: For newer reports, which are available on the web, you can download the original document, which contains the charts. The website to use is on the cover of each report. For older reports, you can currently try third parties, such as the report archive run by Embry-Riddle Aeronautical University/Prescott at amelia.db.erau.edu. Reports are also available on paper from the National Technical Information Service (www.ntis.gov) and General Microfilm (www.general-microfilm.com) for a fee.

Q: I'd like to include my agency's reports in your compilation. How should I proceed?

Q: I was involved in an incident and have insights to provide. May I contribute?

A: Yes. "Reality" often has many different perspectives and rebuttal essays are certainly welcome. We'll figure out a way to put them on the website, once we determine the interest. Please contact us to discuss the particulars: feedback@fss.aero.

Q: I found a technical problem, such as a URL/link not resolving correctly, or a missing page.

Q: I'd like to make copies of a particular report for my class or to hand around. Is that OK?

A. Take a look at the publishing agency's website, or contact us. To stay legal, different countries have different standards on copyright. I think I've sorted through most of them, and can point you in the right direction to get clearance from the investigative agency. In general, it shouldn't be a problem. feedback@fss.aero.

A: In 1990, I tripped across a box of accident reports at a used-book store. In a subsequent discussion with pilots, and particularly Continental (former Frontier) pilot Mark Ingram, I decided there would be value to making this huge pile of documentation available to other pilots in machine-readable format, such that they could carry them around on layovers, etc. In 1993, a 13-year career interregnum at Apple put the project on hold, more or less.

The overriding objective is to improve safety by making people aware of the past, and, more importantly, aware of the pervasive "cascade of failures" that accompanies every crash. While the reports themselves vary in accuracy and perspective, our objective is to present them as objectively and clinically as possible, and to provide mechanisms for people researching specific types of events to be able to find incidents of a similar theme. Incidents and accidents can be hugely educational "case studies" in which students can sharpen their own understanding of aviation.

Over 4500 reports and synopses were reviewed in order to end up with this selection of 1100-odd reports, which will hopefully be interesting and educational for a professional audience.

1. It is important to realize the limitations of using older reports for study. Technology changes, as does the airspace environment, regulations, company cultures, etc. Users are cautioned to always defer to their own company's policies and flight operations manuals as the ultimate authority as to the correct operation of their airplane (read the AAIB report on the British Midlands crash at Kegworth for good reasons why). This material is for informational purposes only.

2. If you're using this data for something major, like a thesis or dissertation, be sure to contact the investigating agency for the final details on the report. Reports often get revised. Keep in mind that paper is still the gold standard, and many agencies publish reports on a "subscription" principle. Moreoever, the "reference" report is often not in English; the English translations are made available by the investigative agency for the larger community. Invariably, imprecision in translation will exist, so the user is again referred to the original report.

It's also important to note that revision notices might get published as part of a later, non-related report, leaving the original report intact with its errors. The agency can usually help make sure you're working with the most recent, accurate facts.

Besides reporting major events, "minor" events in which good airmanship saved the day are also featured, thus demonstrating decision-making processes which broke the "cascade of failure." Hence the huge number of smoke emergencies in the reports.

So there aren't many "ground vehicle collides with airplane" reports (though this is a major problem for the industry). We also wanted to include reports with some meat to them, so even though a one-paragraph "incident" report might look interesting, if there was nothing substantial explaining it, it was a mere factoid and thus data "noise."

For better or worse, the late 1960s started a revolution in flight training and operations, the "need to know" concept, in which the amount of technical information required by pilots was greatly reduced, and procedures-based operations was much more emphasized. This, combined with jet operations, marked a significant difference between airline operations and causal factors involved in older reports and the selection chosen. A smattering of older reports of historical significance are included, though, such as the Grand Canyon midair.

A: Two reasons. First, there is value in seeing the historical documents as they were presented. Second, in an earlier stab at this project a text-only approach was attempted, but there were so many informative artifacts in the report that were difficult to resolve, there was really no choice but to go with the "intact" approach. For example, do we convert/reformat the appendices, which might include graphs and charts, do we impose our own uniform standard for cockpit voice recorder transcripts, etc. Ultimately, if this approach had been taken, it would have been tantamount to would have been republishing this data, which wasn't really the project's objective--the original data was designed for a print audience; our objective is reproducing the data for increased accessibility.

A: Many reports lack a citable title. Most reports have awful, undescriptive titles. Some countries include as much factual information in the title as possible, while others are content with the date and registration # of the airplane. A very functional, just-the-facts title scheme is used, focusing on the type of event (not necessarily the cause), airplane type, and date. In cases where the report already had a title, the original title is retained, but sometimes embellished.

When citing reports, be sure to use the title on the actual report, not the one in the database.

A: The keyword classification scheme evolved as the project evolved. Ultimately, we came up with a "Operations/Systems/Consequences/Other" model.

- Systems focuses on causal failure in systems (as opposed to people), BUT can also be used to classify results. A good example is a landing gear failure following a hard landing. While the hard landing may have produced the failure, the failure of the landing gear produced the evacuation which produced a flight attendant's broken ankle. So the failure of the landing gear is both a result-of and a cause-of an event.

- The Consequences keywords focus on the bottom line result: what happened? Why do we care? Was someone injured? An airplane lost? Damage (notwithstanding the above caveats). With respect to injuries, this project was not going to be a gore-fest, so passenger and pilot injuries are mostly ignored. On the other hand, flight attendant and ground worker injuries are highlighted, since these are not normally covered or summarized in projects like this.

- Lastly, the Other keywords focus on external factors affecting the event, such as regulatory oversight, management or workplace culture, scheduling/fatigue, etc.

Aircraft accident means an occurrence associated with the operation of an aircraft which takes place between the time any person boards the aircraft with the intention of flight and all such persons have disembarked, and in which any person suffers death or serious injury, or in which the aircraft receives substantial damage.

Incident means an occurrence other than an accident, associated with the operation of an aircraft, which affects or could affect the safety of operations.

In our compilation, an "incident" or "event" may be used as an umbrella term including accidents, or sometimes "event". However, note that "incident" and "accident" are precisely defined terms when used within reports.

The definitions above were also used as the primary criteria to keep away from the CHiRP or ASRS databases, by mainly focusing on events of sufficient impact to warrant formal investigations.

Many countries also define a "Serious Incident". The UK Air Accident Investigation Branch, for instance, defines a Serious Incident as an incident involving cricumstances indicating that an accident nearly occurred. They have an excellent summary on:

Again, there are slight differences between countries. The UK Civil Aviation Authority (www.caa.co.uk) defines an "Airprox" (Air Proximity) as follows:

For every Airprox incident assessed the [Airprox] Board determines one or more causal factors. These tell us why events started in each instance and signposts the lessons to emerge. It is self evident that attention paid to 'cause' is worthwhile because it is likely to deliver and promote better prevention'. The UKAB database records record causal factors under 83 separate description.
Risk Categories
Risk level assessments are made on the basis of what actually took place and not on what may or may not have happened. There are four categories, agreed at international level, as follows:
• A Risk of Collision An actual risk of collision existed
• B Safety not assured The safety of the aircraft was compromised
• C No risk of collision No risk of collision existed
• D Risk not determined Insufficient information was available to determine the risk involved, or inconclusive or conflicting evidence precluded such determination

The US Federal Aviation Administration prescribes separation minima, but does not appear to have a similar level of classification for separation loss.

Q: Some of the synopses mention the destination by city, instead of by airport. Yet you list a specific airport, even if there are several candidates in the city. How did you do this?

A: First, I took a look at the city. For example, if it had two airports, and one was an international airport and the other only had a 4000' runway, or had a mandate explicitly for light/corporate traffic (verified through their webpage), I'd choose the international airport. Unless it was an airline charter, which can be ambiguous.

If the case was more ambiguous, such as two international airports, I'd look at the time frame. For example, Tokyo Narita wasn't opened until 1978, so international airline flights to "Tokyo" before then I can assume went to Haneda.

If there were two or more reasonable candidates (such as Paris), I'd look to see if the flight number was listed. Airlines do not tend to change flight numbers, so if the report was recent and I could find scheduling information, this could reliably answer the question.

If there were two or more reasonable candidates and no flight number, I'd look up the airline, based on either its name in the report or the airplane registration for the period in question, then look at the timetable for the airline. If it only flies to one airport in the city, that would be our choice.

Lastly, if there was true uncertainty, such as a business jet flying to a city with multiple candidates and no way to verify the airport's identity, or an airline-charter, or an airline flying to multiple destinations in the city, I'd cite the destination as "Unknown."

A: Since our primary audience is comprised of pilots, 4-digit ICAO codes are used for sanity's sake. A variety of resources are available on the Internet to map these to ATA codes (ATA codes are the traditional 3-digit codes you see on baggage tags, the "passenger interface" to airports).

A: When I started, the idea was to simply indicate what the route was. This can be informative on many levels, e.g., a short-term flight vs. a transoceanic EROPS operation.

Then it became apparent that most reports do not indicate the origin and destination. This caused us to rethink the problem as just a need to list the facts: where the airplane started, and where it ended up.

In the case of a flight-terminating crash, I leave the "scheduled" destination intact and note the wreckage site.

In the case of a crew-guided rethink, e.g., a rejected takeoff terminating the flight, or a gate accident terminating the flight, I also leave the "scheduled" destination intact, if known.

In the case of an in-flight diversion, I list the diversion airport as simply the updated destination. I have included a search function that lists events (almost always successful) in which the crew opted to divert. This can be a valuable learning aid!

A: The specific model is listed, when available. For searching, a keyword is defined that represents that family, e.g., the Douglas DC-9 would cover everything up to but not including the MD-80.

The issue of how to handle "inheritances" was also tricky. No doubt there are explicit current trademarks for many of these, but in the historical context, they may not make much sense. This would apply to, for instance, Hawker Siddley inheriting the de Havilland Trident, or Boeing inheriting McDonnell Dougas, which itself was the result of a merger. So in the latter, is a DC-8 a Douglas DC-8 or a Boeing DC-8? The latter doesn't seem too wholesome, so I went with the former.

The DC-9-80 is referred to as the McDonnell Douglas MD-80, since it was a big project/rebranding after the Douglas/McDonnell merger.

Since the MD-90 was ultimately manufactured after the Boeing/McDonnell Douglas merger (though most design work occurred before), it is listed as Boeing.

Since the scope of this project is mainly jet transports, all helicopter incidents are lumped into a simple family classification of "Helicopters."

For manufacturers which only have a few airplanes, the manufacturer name is used as a Family classification. Some people might view "Cessna" dismissively as a Cessna 172, as the general media does: I do not, and am simply trying to make grappling with this mass of data a bit easier. So "Cessna" includes the whole range of its fine products, from the 150 to the Citation.

We mean no disrespect at all to the trademark owners: as a literary exercise, we're trying to convey a sense of historical perspective.

A: When these old birds crash, it's a loss to our heritage and should be noted. It's outside the scope of the project, but when I noted this, it seemed reasonable to include it.

Q: I don't agree with one of your keyword classifications, or have another one I think might be of value.

Q: You have a billion classifications. Why nothing for spatial disorientation or get-there-itis?

A: I find, in general, that speculating on the perceptive state of a dead pilot and stating that as fact is inappropriate. Clearly, there may be reasons for speculating this as the cause of an event, and they should be listed as Findings. But to specify it as Probable Cause needs to be very carefully framed, which it usually is not. So I try not to feed the "speculation" monster. For this reason, there are no keywords which focus on pilot mental state, such as spatial disorientation or "get-there-itis." These are VERY well-established and very real and very dangerous phenomena and are of concern to any pilot. They just don't have a place here.

A: Since so many of these are unpredictable in nature, are of short duration, and just "happen", I were tempted to drop them from the summary. However, after consulting with several airline pilots and especially Jim Irving, I was convinced that this is a significant enough issue to warrant enhanced vigilance, which is one of the purposes of this project, so they all went back in.

I'm glad I did: some interesting trends are very clear. Interestingly, few passengers seem to be injured, but a TON of flight attendants receive serious injuries, usually a broken ankle, foot, or leg, in that order. One would speculate there should be an "anchor in place" function for serving carts, and F/As should flee to their seats when the belt signs go on. Many of the reports contain information on G-force ranges and time periods.

Copyright © 2006-2009 by Flight Simulation Systems, LLC.  All Rights Reserved.

All referenced trademarks are the property of their respective owners.