This article is part 4 in a series on errors in machine readable travel documents. The previous articles focused on various international travel documents from countries such as Romania, South-Korea and Somalia1. In this part John Mercer takes a closer look at the travel documents of his own country, the United States of America.
The focus of these articles has not been on the occasional typographic error or other one‑time slip‑ups, but on the systemic error, one that was made in the design or production of the document. Although the United States generally gets things right with regard to travel documents, there are several instances where that has not been the case.
Machine readable zone
Many of the instances relate to only a few letters in the machine readable zone (MRZ). However, if even one of the characters is wrong, the holder should be subject to extra scrutiny in inspection, slowing down a process that is already time sensitive. Little differences matter.
The first letter of the first MRZ line is important, because it tells the document reader what document it is, and what format the data will be. The present version of the ICAO Doc 9303 standard for machine readable passports mandates that for book‑form documents, the first letter of the MRZ shall be a ‘P’, which may be interpreted as telling the MRZ reader to expect data in a passport‑like string, two lines of 44 characters each2.
US-Mexico border crossing cards
The land border between the US and Mexico is approximately 1,950 miles or 3,140 km long. Much of this distance follows the course of the Rio Grande River, and a lot of the remainder is arid desert or mountainous terrain. Consequently, much of the legal border crossing is concentrated at passes or roads on land, and bridges over the river.
Illegal immigrants without papers chance inhospitable and often fatal passages through the uninhabited areas, while illegals with fake ID documents will try the monitored border crossings.
To help control its borders, the United States began issuing machine readable visas in 1991, after several years of developmental work. These visa labels were intaglio and offset printed, and personalised on laser toner printers. The installation and operating costs of this process were higher than the previous multicolour visa stamp. Many Mexican citizens were repeat visa customers, and to iteratively verify these citizens and issue visas again was not the best use of scarce consular officer time and resources.
In 1992, to avoid repetitious visa issuance, US Consular Offices in Mexico began issuing the MICA card, which is a stand‑alone B1/B2 visa. This was in the form of a laminated ID‑3 size card (see figure 1). The card and the first machine readable visas were issued in anticipation of the first ICAO standard for machine readable visas, Doc 9303 Part 23.
The error in the MICA card can be found within the first five characters of the MRZ, which are ‘VNMEX’. The ‘V’ for visa and ‘N’ for non‑immigrant are correct, but the next three letters describe the country issuing the card, which should have been ‘USA’, and not ‘MEX’ for Mexico. This usage lasted until the mid‑1990s when the correct letters ‘VNUSA’ came into use for the US visa.
US B1/B2 Border Crossing Card
The US Immigration and Naturalization Service (INS) also issued a border crossing card. Early versions of the I‑586 card were not machine readable, but by 1990 a revised version was issued, using an ‘A’, one of the three correct initial letters A, C or I, for the first position in the MRZ of an ID‑1 card.
In 1996, the US passed a law (the Illegal Immigration Reform and Immigrant Responsibility Act, or IIRIRA) that mandated issuance of a border crossing card that contained a biometric measure describing the holder. The US State Department developed a Border Crossing Card (BCC), based on the then‑current INS Permanent Resident, or Green Card. Exit the MICA, and enter the BCC.
The storage technology used in the Green Card was optical memory based, capable of storing up to 2 MB of data, depending on the storage area. Because of the choice of a very wide optical memory format, the machine readable zone had to be located on the front of the card, sharing the limited ID‑1 space with the necessary eye readable personal and document data. This forced relocation of the MRZ to the front of the card renders the card non‑compliant with the ICAO card specifications, but does not hinder the actual readability of the card.
The type of ID‑1 card is supposed to be indicated by the first character of the MRZ, using as a first letter the characters ‘A’, ‘C’, or ‘I’, and a second character. INS, to their credit, had done this correctly when they first began issuing machine readable Green cards.
As a Projects Officer for the US State Department – Bureau of Consular Affairs, the author was faced with the decision of establishing the MRZ for the new smaller BCC. He reasoned that because the card was a visa, the first letter should be a ‘V’, analogous to the MICA card, and the second letter should be a ‘B’ for Border Crossing. Thus the machine readable zone reasonably started with the letters ‘VB’. The only problem was that it was wrong according to the specification for cards (see figure 2). This use continues to current times, as the most recent version of the Border Crossing Card still shows the use of the incorrect ‘VB’ document type (see figure 3).
The general lesson to be learned from this is that the format of the document, be it card, label or book, will determine the proper document identification code, rather than the actual intended use of the card.
Refugee Travel Document and Re-entry Permit
A second example of the disconnect between the proper identification characters and the document identification code is the case of the United Nations Convention Travel Document (UNCTD). States who are signatories to the UN Conventions on Refugees and Stateless Persons are to issue travel documents to persons who need to travel, but no longer have or can use the passport of their native country.
The UNCTD is in the form of a book, and has pages for visas and entry/exit cachets. The original documents date to 1951 (Convention Relating to the Status of Refugees) and 1954 (Convention Relating to the Status of Stateless Persons), well before machine readable document standards existed.
For many years after the ICAO Doc 9303 machine readable passport standard first appeared in 1981, no provision was made for the UNCTDs, as they were handwritten. But gradually States began to issue machine readable CTDs, inspired by ICAO Doc 9303, but not necessarily in conformance. There was no consensus as to the letters used to designate the type of document. The standard specified two characters, the first a ‘P’ for ‘passport’, and the second a variable letter. Others noted that alternate books were not passports, and different first letters began to be used in the MRZ.
The United States version of the UNCTD documents, the I‑327 Re‑entry Permit and the I‑571 Refugee Travel Document, were first issued around 1993 by the INS. Both documents used ‘TP’ as the document identity letters. As ‘TP’ is not compliant with ICAO Doc 9303, the letters would have caused problems each time the MRZ of sucha document was read at a border crossing post. Figure 4 shows a typical data page for the Refugee Travel Document, with the first two letters being ‘TP’. This document also shows the three‑quarter view portrait in use in these documents prior to the advent of full face images that facilitate facial match algorithms.
For a short period of time, ICAO Doc 9303 allowed for special letters to designate the special travel documents that were book form, but were not passports:
• ‘L’ for the United Nations Laissez Passer.
• ‘R’ for the Refugee/Resident aliens.
• ‘S’ for Seaman’s Identification books.
This provision in the fifth edition of ICAO Doc 9303, footnote m after Section 6.6, was removed from the subsequent sixth edition in 2006. As mentioned above, the present version of ICAO Doc 9303 mandates that for book‑form documents, the first letter of the MRZ shall be a ‘P’.
This creates the delicious irony that the type of document is mandated to be a ‘P’ for ‘passport’, but that there is also stipulated that the legend “This is not a passport” be written on the document, often in the facing page. Very reminiscent of the famous Rene Magritte painting of a pipe, but with the title “Ceci n’est pas une pipe”.Only with the most recent versions of the Refugee Travel Document and Re‑entry Permit has the Department of Homeland Security begun issuing these documents with correct MRZs, using a ‘P’ as first letter, see figure 5.
Check digit miscalculation
The previous examples of US‑issued incorrectly reading MRZ documents were systemic, in that all documents were issued with the same flaw in reading. For the MICA and its successor the BCC it probably did not matter, as they were pretty much valid only to cross the land border between US and Mexico, and only read in US‑owned document readers.
In 1993, the INS began a programme of issuing cards to frequent border crossers that incorporated a hand geometry biometric, which identifies holders by the shape of their hands. Because hand geometry only requires 14 characters to store the data, it was possible to store the hand geometry template in the MRZ on the card. This provided a reasonable identification, and was easy to use. Unfortunately, the level of differentiation between people, about 1 in 700, is much less than that required to identify people uniquely, and so hand geometry was not considered when ICAO began to choose the biometrics and storage media that have led to the present facial and fingerprint images stored on a contactless IC chip.
An example of check digit miscalculation occurred on the author’s own PORTPASS card (to use in Port Passenger Accelerated Service Systems, to facilitate the entry of pre-screened low-risk travellers into the US). When correctly calculated, the date data yields a check digit of ‘5’ after ‘440823’ (see figure 6, top), and in the incorrect case yields a ‘3’ (figure 6, bottom). Because these cards were only used to enter the US and not in international travel, the impact on slowing passenger flow was small.
US passport errors
The US has not had many systemic errors in the MRZs of its passports. The longest running discussion in former times was the check digit for the optional field. ICAO Doc 9303 stipulates that a check digit be used, in the penultimate position in the bottom MRZ line. If there is no data in the optional field, the check digit should have been ‘0’. The US used a filler character ‘<’. This did not matter, as in the calculation of the overall check digit for the MRZ, both a zero ‘0’ and a filler character ‘<’ count as zero, so there is no effective difference.
A second variation from common practice is the location of the signature. Rather than on the data page, the US practice places the signature on the page adjacent to the data page. This is allowed but not encouraged in the ICAO passport standard.
Of course there are human errors and the occasional coding error. However, these are one‑off mistakes, and they do not reflect on the systemic correctness of the process.
US Passport Card
The US Passport card was issued in response to the Western Hemisphere Travel Initiative (WHTI), signed in 2004 with implementation beginning in 2007. The initial assumptions were that the passport card would conform to the passport card specifications in Volume 1 and 2 of the applicable parts of ICAO Doc 9303. However, several changes were made. First the required personal signature on the face of the card was dispensed with, on the grounds that the signature was deemed to no longer be a good security element (see figure 7).
Secondly, the electronic portion of the card was made to a different ISO standard, one used where distance of reading is more important than data carried. This is because the users of the card, officers the Department of Homeland Security, need to know several cars in advance who is driving up to the land border crossings. Thus, a short electronic key that can be read from several car lengths is better than the ICAO compliant ISO 14443 chip carrying 64K or more of data, but with a nominal read range of 10 cm (4 inches). The key is used to retrieve data stored elsewhere in the system concerning the proper holder of the card. The idea is that when the motorist appears at the inspection booth, the border officer already has the passenger data queued up on his monitor, thus saving quite a bit of inspection time per vehicle.
In summary, given the evolving nature of travel document standards and the limitations on validity for particular purposes, it is not surprising that discrepancies would occur between the standards, which can and have changed over time, and the documents made to those standards. Most of the discrepant documents have been withdrawn from circulation, or replaced by documents that are newer and generally more conformant to standards. However, it is instructive to examine the reasons why document design decisions are made, and why documents are issued that do not conform to international standards. This concern applies to all nations and documents that are regularly used in crossing international borders.
1 Mercer, J. Errors in travel documents, Part 1, 2 and 3. Keesing Journal of Documents & Identity, Issues 26 (2008), 27 (2008) & 34 (2011) respectively.
2 International Civil Aviation Organization (ICAO), Doc 9303: Machine Readable Travel Documents, Part 1: Machine Readable Passports, current version: sixth edition 2006; http://www.icao.int/publications/Documents/9303_p1_v1_cons_en.pdf
3 International Civil Aviation Organization (ICAO), Doc 9303: Machine Readable Travel Documents, Part 2: Machine Readable Visas, current version: third edition 2005;