On March 12, 1948, at 9:14 p.m. Pacific Standard Time, Northwest Airlines Flight 4422 crashed into Mount Sanford, a peak in the remote Wrangell Mountains in eastern Alaska. All 24 passengers—merchant mariners returning to the U.S. from Shanghai, China—along with six Northwest crew members, probably died on impact. The debris, too difficult to reach, was quickly covered by snow and eventually entombed by ice.
There it remained until 1999, despite many failed efforts to find it. In that year Kevin McGregor and Marc Millican, two former U.S. Air Force pilots who like to solve forgotten aviation mysteries, having determined that the glacier containing the plane was retreating, gained permission from the National Park Service to recover parts of the wreckage if they could find it. After an arduous climb, they discovered scattered debris, along with a desiccated left arm and attached hand in the ice. As McGregor explains, “That changed the entire project. We became compelled to find out to whom the arm and hand belonged.”
McGregor and Millican’s quest to identify the remains eventually led to an unusual collaboration of DNA experts, fingerprint analysts and forensic genealogists, including myself. Our challenging and ultimately successful effort may benefit many more families than just those of the doomed men onboard the Northwest flight. Some of the laboratory techniques developed during our investigation may one day prove helpful in identifying victims of mass disasters and more than 800 unknown soldiers who died during the Korean War.
The discovery of human remains brought Alaska law-enforcement agencies into the picture. A state trooper carefully freed the arm and transported it to the medical examiner’s office in Anchorage, some 200 miles away. After taking impressions of the finger pads, the medical examiner embalmed the remains.
Because the arm and hand bore no distinguishing marks, fingerprinting and DNA analysis were the only possibilities for making a conclusive identification. After three years, however, it became clear that standard methods would not solve the mystery. An extensive search for fingerprint records turned up only 22 so-called ten-print cards, leaving no records for eight victims, including the six crew members. Even had there been a full set of reference prints for each victim, however, the dried-out hand was so badly damaged by exposure to the elements that the Alaska medical examiner’s office would have been unable to make a positive identification.
Investigators were stymied on the DNA front as well. In 2002 the medical examiner’s office sent a tissue sample to a commercial DNA laboratory. Alas, the lab reported that the biological material had “degraded to a point where the DNA strands were too small to get intelligible results.”
McGregor and Millican—now joined by Randall Haslett, the son of the flight’s purser—decided to search for a research scientist who specialized in making identifications with ancient DNA. Their quest led first to Ryan Parr of Genesis Genomics (now Mitomics) and then, in 2006, to Odile Loreille of the Armed Forces DNA Identification Laboratory (AFDIL) in Rockville, Md. Loreille is known for analyzing highly degraded DNA. Rather than looking at the DNA found in the nuclei of cells, however, she studies the DNA in mitochondria, the tiny organelles that cells use to create energy. Because cells have so much more mitochondrial DNA (mtDNA) than nuclear DNA, mtDNA offers a better chance of identifying very degraded remains.
Loreille was interested in the Northwest Airlines project because she thought it might help her solve the mystery that brought her to AFDIL: how to identify the remains of more than 800 unidentified U.S. soldiers from the Korean War. Most of these men are interred in Hawaii in the National Memorial Cemetery of the Pacific, otherwise known as the Punchbowl. The formaldehyde used to embalm the servicemen’s remains had substantially damaged their DNA. If Loreille could use the Northwest case to develop new techniques for analyzing embalmed tissue, it would be another step in her efforts to help the armed forces identify the remains of these Korean War veterans.
All in all, Loreille knew, the best chance of success was to obtain DNA from the arm’s bone tissue, which is usually better protected from contamination by the environment and from the DNA of anyone who handled the remains. She had recently discovered how to more efficiently separate formaldehyde residue from bone. But even that process was unlikely to generate enough material. During the course of our investigation, however, Loreille developed a demineralization process that completely dissolved the bone matrix, providing just enough mtDNA for analysis.
Of course, DNA extraction was only half of the story. To make an identification, the mtDNA from the decades-old tissue would have to be compared with that of a family reference for each candidate until a match was obtained. Because mtDNA is passed to each child only from the mother, any male or female relative could serve as a reference as long as he or she was linked to the candidate through an exclusively matrilineal line. This requirement often makes it difficult to locate distant relatives who can provide mtDNA, given that a woman’s family name typically changes at marriage. That is where I came in. As a forensic genealogist, I have traced hundreds of people worldwide for many reasons, including DNA referencing for the military and in connection with historical projects.
In trying to narrow the possibilities while speeding up the identification process, Loreille turned for help in 2007 to Ted Robinson, an assistant professor of forensic science at George Washington University. Although the earlier fingerprint analysis had not been successful, a second attempt with new techniques might con-
ceivably rule out some of the candidates. Then it would not be necessary to locate living relatives to provide DNA references for all 30 men.
The fingerprint analysis, performed in parallel with the mtDNA identification, quickly presented its own challenges. Fingerprint identification relies on three levels of detail. Level 1 takes into account the general pattern of the skin’s friction ridges, which allow an individual to grip objects. This pattern falls into one of three categories—loops, whorls or arches. (There is only one type of ridge pattern per finger.) Level 2 details are known as minutiae, or Galton points, in honor of Sir Francis Galton, whose work laid the foundation in 1892 for the current system of recording and identifying fingerprints. Minutiae include places where the line of an individual ridge splits into two, develops a spur, includes a dot or simply comes to an end. On the finest scale, level 3 describes the characteristics of individual ridges, such as their thickness and their level of convexity or concavity. It also includes the locations of sweat pores. Comparing level 1 details in two sets of fingerprints is sufficient to rule someone out as a match; however, it is not specific enough to use for a positive identification. Level 2 and 3 details must be used for such determinations.
By this point, the epidermal layer of skin was no longer present on the fingers—it had sloughed off since the arm’s removal from the ice, and the underlying dermis was almost smooth. Furthermore, only 16 of the original 22 ten-print cards from Northwest Flight 4422 could now be located, so there were no reference prints for 14 of the victims. Nevertheless, Robinson persevered. He attempted to restore the pliability of the skin by bathing it in specially formulated rehydration fluids. Forensic scientists refer to this process as fingerprint rejuvenation. Coincidentally, Robinson had just met Michael Grimm, a retired supervisor from the state of Virginia’s Department of Forensic Science, at a conference. Grimm gave Robinson a sample of a new rejuvenation fluid that had been employed in the identification of victims of Hurricane Katrina in 2005 and that could possibly produce results within hours.
The fluid did the trick. Robinson soaked the hand at 122 degrees Fahrenheit (50 degrees Celsius), checking the results hourly as finger-ridge detail slowly emerged on all five fingers. After photographing the results, Robinson took casts of the prints using two types of silicone rubber. When he removed the finger pads and soaked them separately, the fingerprint detail improved even more. After Grimm photographed the finger pad casts and imported the digital images into a photo-enhancing software program, the prints were so clear that even the sweat pores were visible on the 60-year-old hand.
Ironically, the high quality of the photographs created a new problem. As Robinson explains, “A lot of [the mariners’ cards] were overinked, smudged and just poorly done. The fingerprints from the hand were now better than the prints from the ten-print cards. Because of the poor quality of the ten-print cards, identification could not be made.” Still, Robinson knew that all five fingers of the hand from the crash site had loops, so he was able to rule out 10 victims by discerning that each had at least one finger with an arch or a whorl on his left hand. Grimm eliminated four more based on the finer details in the loops.
Identification at Last
In the meantime, forensic genealogist Chriss Lyon and I worked to find living relatives who might provide the necessary reference samples for the remaining victims. By September 2007, 13 of the 30 men had been ruled out by mtDNA, nine had been ruled out on the basis of fingerprints alone, and five had been ruled out by both mtDNA and fingerprints. That left three men: purser Robert Haslett and merchant mariners Francis “Frank” Joseph van Zandt and John V. Elkins.
Unfortunately, the fingerprint records for Haslett were illegible, and there were no living matrilineal relatives for him who could provide mitochondrial DNA for comparison. (A mitochondrial DNA sample from Haslett’s son, Randall, could prove a relationship only to Randall’s mother and her relatives.) But father and son did, of course, share their Y chromosome, so Loreille used advanced laboratory techniques to amplify the amount of DNA from the nuclear material in the arm to create a partial profile of the unknown victim’s Y DNA. It did not match Randall’s.
Only two candidates remained. John Elkins’s relatives declined to give samples of their DNA, and his fingerprint records were too smudged to be of use. Fortunately, our luck was about to change, finally giving our team what we needed to determine whether the arm had belonged to Elkins or van Zandt.
According to vital records, Frank van Zandt was born on October 21, 1911, in Bennington, Vt., the youngest child of Orville van Zandt, Sr., from New York State and Margaret Conway from Ireland. Frank had one sister, named Elizabeth (whose children might have served as mtDNA references), but we found no trace of her or any possible descendants after the 1910 U.S. census. Going back a generation, the search for collateral female-linked Conway lines in the U.S. also ran into trouble. I learned that Margaret Conway immigrated with two sisters (and three brothers) to the U.S. in the 1890s. Unfortunately, one sister never married, and the other sister did not have any surviving female lines.
Perhaps Margaret had left sisters behind in Ireland? I had done 40 years of Irish genealogical research, and so I knew that Irish civil and church records are organized by county. To find Margaret, I had to discover her county of origin. After searching through thousands of records, I got a lucky break from Bill Budde, the archivist at the Bennington Museum. Budde discovered that the 1936 marriage record of Frank’s brother, Orville, Jr., recorded his (and therefore Frank’s) mother’s birthplace as “Limerick.” A search of Irish birth registrations revealed that Margaret was born September 14, 1871, to John Conway and Ellen Drumm from County Limerick. There was more good news: Margaret had left three sisters and a brother in Ireland. But finding their descendants more than 100 years later was not going to be easy.
During my painstaking search for Conway-Drumm descendants, I was eventually referred to Maurice Conway, the patriarch of the Conway family of the village of Askeaton. He did not initially recognize any of the names we had of van Zandt’s Conway ancestors. Ultimately, however, I learned that Maurice’s maternal great-great-grandmother, Elizabeth, was Ellen Drumm’s sister—that is to say, he and van Zandt shared a common maternal ancestor. Because Maurice was a matrilineal relative, therefore, a sample of his mitochondrial DNA could be used for the identification.
Loreille compared the mtDNA from the arm against all 19 of the reference samples that were now available for the men onboard Flight 4422. The DNA sequence from the remains matched one reference only, that of van Zandt’s maternal cousin, Maurice Conway. For added confirmation, we located van Zandt’s brother’s son, who agreed to serve as a Y-DNA reference. The partial Y-DNA profile that had ruled out Robert Haslett matched that of van Zandt’s nephew at every locus.
Robinson and Grimm had meanwhile discovered that in their search for fingerprint records, they had been asking the wrong question. Instead of requesting “official” fingerprint records, they should have been asking for “any” fingerprint records. They were surprised to learn that the National Maritime Center had extra fingerprint records of many of the merchant mariners that had been taken when they signed on to a new ship. These new records gave us van Zandt’s prints for the first time, allowing Robinson and Grimm to match van Zandt’s fingerprints to those taken from the 60-year-old hand. Their efforts produced the oldest postmortem fingerprint identification on record.
We now had independent, corroborating results from both DNA and fingerprint analysis identifying the arm in the ice as having belonged to Francis Joseph van Zandt. As for the unknown soldiers of the Korean War, Loreille continues her research to identify their remains. Her work on extracting DNA from embalmed tissue suggests that it may be possible to recover enough mtDNA from the Korean-era remains to identify them. She is now working with newly developed DNA-sequencing technologies that, in the next few years, might make identifications feasible using extremely small amounts of DNA—whether from long-dead soldiers or victims of mass disasters.
Our results also showed the importance of working across disciplines. Whereas DNA experts, fingerprint analysts and forensic genealogists often try to answer the same questions about identity and relationships, we typically confine our efforts to our respective professional domains. Our highly collaborative investigation of Northwest Flight 4422 shows that cross-disciplinary efforts can produce robust results, especially in very difficult cases.