The Sperry rail car was the detector car that started commercial internal rail testing in the United States, in November 1928. It found hidden flaws with electromagnetic induction, not sound. Ultrasonic testing came later: by hand from about 1949, and on test cars from about 1959 to 1960.
Federal law followed decades after the technology. FRA's first Track Safety Standards, published on October 20, 1971, required a yearly "continuous search for internal defects" on higher-speed track. That requirement, now 49 CFR 213.237, is the reason rail flaw testing is a regular, scheduled job today.
This article sets out the dated history, separates what the records confirm from what is often repeated, and explains how each step shaped the way rail is tested now.
Why did American railroads need a rail detector car?
Railroads needed a detector car because the defect that broke rails most often, the transverse fissure, grows inside the rail head where no inspector can see it. Before detector cars, the only defenses were walking inspections and removing whole batches of suspect rail.
Federal accident reports of the period make the problem plain. In its report on a 1930 derailment near Munfordville, Kentucky, the Interstate Commerce Commission (ICC) recorded a track supervisor's view that transverse fissures in 100 lb rail were "barely discernible" even after they reached the surface (ICC report, Louisville & Nashville, Oct 3, 1930, ROSA P).
The same report describes the older defense. When fissures developed in two rails from the same steel heat, the railroad removed rails of that heat from the high side of curves, and when four developed, it removed every rail from that heat. A 1955 Western Pacific employee magazine summed up the pre-detector era the same way: railroads could guard against defective rail "only by mass and speculative rail replacement programs", which wasted good rail and still missed bad rail (WP Mileposts, June 1955).
Key takeaway: the detector car replaced guesswork about whole heats of rail with a test of each rail in track.
Did the 1911 Manchester wreck lead to the Sperry rail car?
The 1911 Manchester wreck is a real, documented transverse fissure failure, but the record does not show that it directly started Elmer Sperry's work. It is better described as part of the background that made internal rail defects a national safety issue.
On August 25, 1911, Lehigh Valley Railroad train No. 4 derailed near Manchester, New York, when a rail broke. Most accounts give 29 deaths and about 60 injuries. A 1913 ICC report on a different derailment compares its broken rail with "the broken rail that caused the accident on the Lehigh Valley Railroad at Manchester, N.Y., on August 25, 1911", and notes that the Manchester rail had transverse fissures, was piped, and had longitudinal seams in the head (ICC report, Hays Mill, Ala., Aug 15, 1913, ROSA P).
The gap is in the dates. A 1999 university transportation report places the start of Sperry's rail work in 1923, and the contract with the American Railway Association came in 1927 (Jeffrey and Peterson, MPC 99-106, 1999). Several later histories open the Sperry story with Manchester, but none of the sources checked for this article ties the two with a document from the time, so treat the link as narrative, not proven cause.
| Claim often repeated | What the checked record supports | Confidence |
|---|---|---|
| A 1911 Manchester, NY wreck killed 29 | Lehigh Valley train No. 4, Aug 25, 1911; 29 deaths in most accounts | High (date and rail failure); Med (casualty count) |
| The Manchester rail failed from a transverse fissure | ICC 1913 report: transverse fissures, plus piping and head seams | High |
| Manchester caused Sperry to build his car | Not shown; Sperry's documented work starts about 1923 and the car in 1927 | Unverified |
Key takeaway: cite Manchester as an early documented transverse fissure wreck, not as the proven trigger for the Sperry rail car.
How did the first Sperry rail car find flaws?
The first Sperry rail car used electromagnetic induction: it passed a heavy electric current through the rail and used search coils to detect the disturbance a flaw caused in the surrounding magnetic field. No sound waves were involved.
The 1999 report describes it this way: "A heavy current was induced through the rail to be tested. Search coils were moved through the resulting magnetic field to find perturbations in this field caused by defects" (MPC 99-106). A rail fissure detector patent assigned to Sperry Products, filed in 1930 by Sperry's engineer Harcourt C. Drake, shows the same principle, with detector coils held at a fixed distance from the rail on a moving carriage (US 1,907,755).
How the first car was developed and accepted
The order of events, as given in the sources checked, runs like this:
- About 1923: Sperry begins work on locating internal rail defects (MPC 99-106).
- 1927: Sperry contracts with the American Railway Association for a test car. Car 101 is built from June and delivered in December 1927. Early tests used a hand-held search unit, which struggled on uneven rail surfaces.
- 1928: the team adopts the induction method. After a test run on September 27, 1928, the ARA rail committee accepts the car on October 2, 1928.
- November 1928: first commercial test, on the Wabash Railway between Montpelier, Ohio, and Clarke Junction, Indiana. Car SRS 102 flagged a transverse fissure, and the rail was later broken open at that point to confirm it.
The development and acceptance dates in steps 2 and 3 come from secondary histories, mainly the Wikipedia summary of Sperry Rail Service, which cites company and railroad histories. The company's own history gives 1928 as its founding year. Other accounts say the inspection service company was set up in Chicago in 1929, so the founding is best written as 1928 to 1929.
The American Railway Association reportedly called the detector "one of the most important safety moves in years" (National Inventors Hall of Fame, Elmer Sperry). Elmer Sperry died on June 16, 1930, less than two years after the first commercial test.
What the induction cars found
A 1941 Railway Gazette article, summarized in the TRB database, reviewed ten years of US detector car service (TRID record 6865). Over five years, about 40% of the defects found were transverse or compound fissures, 40% vertical split heads, 15% horizontal split heads and 5% other types. By then the rails were magnetically energized before a four-coil search unit passed over them, and a "Type 80" upgrade added recording pens to help tell defect types apart.
Key takeaway: the first Sperry rail car was an induction car; it proved that rail could be searched for internal flaws from a moving vehicle.
When did ultrasonic testing enter US rail testing?
Ultrasonic testing entered US rail testing in stages: portable ultrasonic testing from about 1949, and mobile ultrasonic equipment on test cars from about 1959 to 1960, added to the induction cars rather than replacing them. The exact years vary between sources, so they are given as approximate here.
The enabling invention was Floyd Firestone's pulse-echo flaw detector. His patent, "Flaw Detecting Device and Measuring Instrument", was filed on May 27, 1940, and granted on April 21, 1942 (US 2,280,226). It describes sending short bursts of high-frequency vibration into a part and timing the echoes from internal flaws; histories of the instrument say Sperry Products later made it as the Reflectoscope. The broader history is on our page on the history of non-destructive testing.
| Date | Ultrasonic milestone in US rail testing | Source | Confidence |
|---|---|---|---|
| 1940–1942 | Firestone's pulse-echo patent filed and granted | US 2,280,226 | High |
| 1949 | Sperry Rail Service introduces ultrasonic flaw detection, used by hand at joints and other places induction could not test | Wikipedia (rail inspection); secondary histories | Med |
| 1959 | First ultrasonic test car, built for the New York City Transit Authority | MPC 99-106 (1999), citing Clarke (1997) | Med |
| 1960 | Semi-automated mobile ultrasonic equipment added to the induction system | Wikipedia (Sperry Rail Service) | Low–Med |
| 1999 | Ultrasonic is "the predominant rail inspection technology used in North America" | MPC 99-106 | High (as a 1999 statement) |
Why ultrasonics were combined with induction
Induction and ultrasonics see different things, so for decades the larger railbound cars carried both. Secondary histories say induction worked poorly at rail ends, where joint bars and bolts hold rails together, and at switch points, so those places were tested by hand with ultrasonic instruments. That fits the 1971 rule, which required equipment "capable of detecting defects between joint bars".
By 1999 the balance had shifted. The MPC report says most major contractors relied on ultrasonic cars alone, while "some companies do employ larger rail bound units that have ultrasonic and magnetic induction technologies" (MPC 99-106). For the physics behind the method, see how ultrasonic testing works for railways.
Key takeaway: ultrasonics joined the Sperry rail car as an addition to induction, first by hand around 1949 and on test cars around 1959 to 1960.
Timeline of US rail flaw detection, 1911 to 2020
The short version: induction cars from 1928, ultrasonic testing from about 1949, a federal testing rule from 1971, and a performance-based rule from 2014. The table gives each step with its source, so a reader can check any date.
| Year | Event | Source | Confidence |
|---|---|---|---|
| 1911 | Aug 25: Lehigh Valley train No. 4 derails at Manchester, NY; the broken rail had transverse fissures | ICC report, 1913 | High |
| About 1923 | Elmer Sperry begins work on internal rail defect detection | MPC 99-106 | Med |
| 1927 | Contract with the American Railway Association; car 101 built and delivered | Secondary histories | Med |
| 1928 | Induction method adopted; car accepted Oct 2; first commercial test on the Wabash in November | Secondary histories; MPC 99-106 | Med–High |
| 1928–1929 | Sperry Rail Service set up as an inspection service, not a car seller | Company history; secondary accounts | Med |
| 1929 | Dec: a Sperry car tests the L&N Louisville Division and finds 15 fissured rails between mileposts 5 and 113 | ICC report, 1930 | High |
| 1930s | Fleet grows to about ten cars (sources disagree on whether by 1930 or by the end of the decade) | MPC 99-106; Wikipedia | Med |
| 1940–1942 | Firestone pulse-echo ultrasonic patent | US 2,280,226 | High |
| 1941 | Railway Gazette reviews ten years of US detector car service | TRID 6865 (abstract) | Med |
| About 1949 | Ultrasonic flaw detection introduced in rail testing | Secondary histories | Med |
| 1955 | 17 Sperry cars survey about 160,000 miles of track a year, testing at up to 12 mph | WP Mileposts, June 1955 (railroad magazine) | Med |
| 1959–1960 | First mobile ultrasonic test equipment on cars | MPC 99-106; Wikipedia | Low–Med |
| 1970 | Federal Railroad Safety Act of 1970 | 36 FR 20336 | High |
| 1971 | Oct 20: FRA Track Safety Standards (Part 213) published, effective Dec 1, 1971, including 213.237 | 36 FR 20336 | High |
| 1998–1999 | 213.237 revised: 40 MGT or once a year, whichever is shorter (applicable Jan 1, 1999) | CFR 2011 edition | High |
| 2014 | Jan 24: "Improving Rail Integrity" final rule, effective Mar 25, 2014 | 79 FR 4234 | High |
| 2020 | Oct 7: continuous rail testing option added as 213.240 | 85 FR 63362 | Med–High |
The rail accident side of this history is covered in famous train accidents that changed history.
How did federal rules make internal rail testing mandatory?
Internal rail testing became a federal requirement with FRA's Track Safety Standards, published on October 20, 1971, under the Federal Railroad Safety Act of 1970. Before then, testing was a railroad's own choice, bought as a service.
The original 213.237(a) read:
"In addition to the track inspections required by § 213.233, at least once a year a continuous search for internal defects must be made of all jointed and welded rails in classes 4 through 6 track, and class 3 track over which passenger trains operate."
— FRA Track Safety Standards, 36 FR 20336, 20342–43 (Oct 20, 1971)
The 1971 preamble is candid about what the technology of the day could do. FRA deleted the words "and in welds" from the equipment paragraph "because of the limited capability of present inspection devices to detect defects in welds". It kept the requirement that equipment detect defects "between joint bars, in the area enclosed by joint bars".
How the rule changed after 1971
| Period | Internal rail testing requirement (213.237) | Source |
|---|---|---|
| 1971 to 1998 | At least once a year on Class 4–6 track and on Class 3 track with passenger trains | 36 FR 20336 (1971) |
| 1999 to 2014 | Class 4–5, and Class 3 with passenger trains: every 40 MGT or once a year, whichever is shorter. Other Class 3: every 30 MGT or once a year, whichever is longer | 49 CFR 213.237, 2011 edition, govinfo |
| 2014 to now | Testing must keep service failure rates under set limits; Class 4–5, and Class 3 with scheduled passenger trains or a hazmat route: no more than 370 days or 30 MGT apart, whichever is shorter; other Class 3 has a 5-year cap | 79 FR 4234 (Jan 24, 2014); 49 CFR 213.237 |
The 2014 "Improving Rail Integrity" rule, mandated by section 403 of the Rail Safety Improvement Act of 2008, also defined a valid search, added the qualified-operator rule in 49 CFR 213.238 (including at least 160 hours of supervised experience), and required testing of plug rail. In 2020 FRA added continuous rail testing as an option in 49 CFR 213.240. The current rule is explained in full in our guide to 49 CFR 213.237 rail flaw detection requirements.
Key takeaway: federal law made internal rail testing a scheduled duty in 1971, then tied the schedule to tonnage in 1999 and to measured rail failures in 2014.
Worked example: how the testing interval tightened
The same busy line would be tested about once a year under the 1971 rule, about every 324 days under the 1999 rule, and about every 243 days under today's rule. This example shows how the history affects a real testing schedule.
Inputs: a Class 4 freight main line carrying 45 million gross tons (MGT) a year, tested at the regulatory maximum interval. The current-rule row ignores any extra testing needed to meet the service failure limits.
| Rule | Interval limits | Which limit controls at 45 MGT a year | Maximum interval | Tests per year |
|---|---|---|---|---|
| 1971 | Once a year (no tonnage limit) | Calendar | 365 days | 1.0 |
| 1999–2014 | 40 MGT or one year, whichever is shorter | Tonnage: 40 ÷ 45 = 0.889 year | 0.889 × 365 ≈ 324 days | 45 ÷ 40 ≈ 1.13 |
| 2014–now | 30 MGT or 370 days, whichever is shorter | Tonnage: 30 ÷ 45 = 0.667 year | 0.667 × 365 ≈ 243 days | 45 ÷ 30 = 1.5 |
The arithmetic: at 45 MGT a year the line reaches 40 MGT in 40 ÷ 45 of a year, and 30 MGT in 30 ÷ 45 of a year. Both are shorter than the calendar limits, so tonnage controls under the two later rules.
A second example from the record: early defect rates
The 1955 Western Pacific article gives a rare published result from a single test run. Over 709 miles of track, the car found 33 transverse defects, which the magazine gave as 4.65 per 100 miles. Recomputed: 33 ÷ 709 × 100 = 4.65 transverse defects per 100 miles.
The same article says 17 cars surveyed about 160,000 miles a year, or about 160,000 ÷ 17 ≈ 9,400 miles per car per year. These are figures the company gave a railroad magazine in 1955, not audited statistics.
Key takeaway: on heavy-tonnage track, today's rule means about one and a half tests a year at the minimum, against one test a year in 1971.
What does a modern combined rail testing system look like?
A modern rail testing system uses several ultrasonic probes at different angles, sometimes alongside induction or other sensors, with suspect locations confirmed by a qualified person on the ground. The basic split between machine detection and human verification dates back to the first cars, which stopped so that a crew could check each indication.
The 1999 MPC report describes the standard ultrasonic layout of its time. Transducers sit in fluid-filled wheels that roll on the rail with couplant; there are three to six transducers in each wheel and usually two wheels per rail, for up to 12 per rail; and the beams are typically 0°, 45° and 70°, plus a "side looker". The cars could test at up to 30 mph non-stop, but "with the stopping and verification rail inspection vehicles typically average 6 to 8 mph in practice".
| Era | Main method | How it was carried | How indications were handled |
|---|---|---|---|
| 1928 to 1940s | Induction (current through the rail, search coils) | Railbound detector car | Paper recorder; car stopped and the crew checked on the ground |
| About 1949 to 1950s | Induction on cars, plus ultrasonic by hand at joints | Car plus portable instruments | Hand ultrasonic at joints and switches |
| About 1959 to 1990s | Induction and ultrasonic on the same car, then ultrasonic-only cars | Railbound cars and road-rail vehicles | Stop-and-verify with a hand flaw detector |
| 2014 to now | Multi-angle ultrasonic (0° for horizontal flaws, 70° for transverse defects), with other sensors on some systems | Railbound cars, road-rail vehicles, walking carts | Valid search under 213.237; qualified operator under 213.238; continuous testing option under 213.240 |
What changed with continuous testing
Under 213.240, a vehicle can test without stopping, and suspect locations are verified later. Suspects must be field-verified within 84 hours, or within 36 hours for the most serious suspected defects, and "a suspect location is not considered a defect under § 213.113(c) until it has been field-verified by a person qualified under § 213.238" (49 CFR 213.240(e)(4)). That rule writes into law the human check the first detector car crews did by hand.
For how a test run works step by step, see our ultrasonic rail flaw detection hub and the explainer on how ultrasonic rail flaw detection works.
Key takeaway: modern systems are faster and record more data, but they still depend on a qualified person confirming each suspect location.
What the history shows about the limits of rail flaw detection
From the first induction car on, rail flaw detection has found most internal defects, not all of them, and a defect can still grow and break between tests. The historical record is unusually clear on this point.
In the 1930 Munfordville case, a Sperry car had tested the line in December 1929 and found 15 fissured rails, but did not flag the rail that broke ten months later. The ICC concluded that the fissure "developed after the Sperry tests were completed" and added a line that still frames how test intervals are set:
"It has not been determined, however, how frequently such examinations must be made to provide immunity from disastrous failures of this character."
— Interstate Commerce Commission, report on the Louisville & Nashville derailment near Munfordville, Ky., October 25, 1930 (ROSA P)
Modern limits, from published test-track and research data:
- Detection is not complete. In a 1999 TTCI evaluation, six hi-rail systems found 72% of 49 known defects; transverse defects under shelling were found about 62% of the time, against 77% for those not under shelling (TTCI TD-99-006).
- Surface damage masks internal flaws. Shelling, head checks and other surface damage can block the ultrasonic beam from defects below.
- Some areas are hard to test. The 1971 preamble already noted weak weld detection; today, thermite weld bases and rail hidden under crossings remain difficult places to search by vehicle.
- A lost test is untested rail. Where coupling is lost or the search is otherwise invalid, 213.237 treats that rail as not tested, and the owner must re-test, slow trains to 25 mph, or replace the rail before the deadline.
- Old figures are not current performance. The 1941 defect mix, the 1955 mileage and the 1999 speed figures describe their own time and equipment; they shouldn't be quoted as how today's systems perform.
How EMA Industries applies this
EMA Industries is a Southern California NDT contractor that provides ultrasonic rail flaw testing for transit agencies, short lines and industrial track owners, and mobilizes nationwide. The method descends from the one described above: multi-angle ultrasonic testing of rail in track, where each suspect location should be verified by a qualified person before the owner decides on remedial action under 213.113. EMA also inspects rail welds, the area the 1971 rule admitted early equipment could not reliably test.
For a track owner, the practical lesson from this history is to plan testing around the 213.237 interval and the rail's own failure record, not only the calendar. To discuss a rail testing program, see our ultrasonic examination services or contact EMA Industries.
FAQs
When did ultrasonic rail testing start in the United States?
Ultrasonic flaw detection entered US rail testing around 1949, first as a portable method used at joints and other places the induction cars could not test well. Mobile ultrasonic equipment on test cars followed around 1959 to 1960. Sources differ by a year or so on each date.
What did the first Sperry rail car detect?
The first Sperry rail car was built to find transverse fissures, internal cracks across the rail head. It used electromagnetic induction, not ultrasound. Induction cars also found split heads and other internal defects; a 1941 review put vertical split heads at about 40% of the defects found.
Did the 1911 Manchester wreck cause the Sperry detector car to be built?
Not on the evidence checked for this article. The 1911 Lehigh Valley wreck at Manchester, New York, is documented as a transverse fissure failure, but Sperry's rail work is dated from about 1923 and the car from 1927, and no record from the time links the two directly.
When did federal law first require internal rail testing?
FRA's Track Safety Standards, published on October 20, 1971, and effective December 1, 1971, first required a yearly continuous search for internal defects on Class 4 to 6 track and on Class 3 track with passenger trains. Today's version is 49 CFR 213.237, last rewritten by the 2014 rail integrity rule.
Is induction testing still used on rail?
Induction is still carried on some larger railbound test cars alongside ultrasonic probes, according to the 1999 MPC report and later descriptions. Most rail testing in North America is ultrasonic, and federal rules don't specify the method; they require that the search be valid and that the operator be qualified under 213.238.
Sources
Regulations and federal records
- FRA Track Safety Standards, 36 FR 20336 (Oct 20, 1971), Federal Register on govinfo
- 49 CFR 213.237, 2011 edition (pre-2014 text), govinfo
- Track Safety Standards; Improving Rail Integrity, 79 FR 4234 (Jan 24, 2014), govinfo
- 49 CFR 213.237 Inspection of rail, eCFR
- 49 CFR 213.238 Qualified operator, eCFR
- 49 CFR 213.240 Continuous rail testing, eCFR
- ICC report on the Louisville & Nashville derailment near Hays Mill, Ala. (Aug 15, 1913), ROSA P
- ICC report on the Louisville & Nashville derailment near Munfordville, Ky. (Oct 25, 1930), ROSA P
Research and history
- Jeffrey, B. D. and Peterson, M. L., Assessment of Rail Flaw Inspection Data, MPC 99-106, Mountain-Plains Consortium, August 1999
- Recent Progress in Rail Fissure Detection, Railway Gazette vol. 75, Aug 29, 1941 (TRID abstract)
- Western Pacific Mileposts, June 1955 (railroad employee magazine)
- TTCI Technology Digest TD-99-006, Garcia and Reiff, February 1999 (hi-rail rail flaw system evaluation)
- US Patent 2,280,226, F. A. Firestone, Flaw Detecting Device and Measuring Instrument (1942)
- US Patent 1,907,755, H. C. Drake, Rail Fissure Detector (1933)
- National Inventors Hall of Fame: Elmer Ambrose Sperry
- Wikipedia: Sperry Rail Service (overview; used for development dates only)
- Wikipedia: Rail inspection, history section (the 1949 ultrasonic date; the section carries no inline citation)