One hundred years ago someone recorded a temperature in Death Valley of 134 degrees.
How reliable is this reading?
Very. Consistently making glass tubes was pretty well developed by that time, and calibrating a well-fabricated thermometer is one of the easiest and most physically reliable jobs in the business.
On the other hand aren’t there strict rules about locating a thermometer to get an accurate reading?
And for the record, the 134 degree reading was made by the National Weather Service, which has had a station there since 1911.
about as accurate as a 1913 ruler.
tube making and thermometers were well established methods…
in the USA weather records through the government started about the mid 1800s. i saw a photo of the site for the record back from then, the thermometer shelter is the style still in use. it is placed about 4 feet above ground and away from obstructions.
On the other hand, not everybody was so careful about siting the thermometer, or even reading it. For a long time the world record highest temperature was thought to be held by a place called Al Aziziyah in Libya, in September 1922, at 58C (136.4F).
However, that was recently wiped from the record books after investigations revealed a number of discrepancies, not only with the siting of the weather station, but also quite possibly by an inexperienced observer reading the wrong end of the maximum marker on the thermometer.
This (and this) is quite an interesting read on how the record came to be overturned.
The same author also raises some questions about the 1913 Death Valley record: link here.
Just to add, that last article says the highest* undisputed* reading (using modern instrumentation) was 129°F (53.9°C) on July 7, 2007, July 20, 2005, and July 18, 1998 at Furnace Creek.
That seems to have been equalled on June 30, 2013, although I can’t find the final official reading anywhere. (There was a 129.9 but that was the National Park Service thermometer, not the official one.)
Former temperature metrologist here. 
LiG thermometers and PRTs built in the late 19th century were capable of high degrees of precision, even by today’s standards. I don’t have a cite, but I would wager the laboratory-grade LIGs and Callendar-type PRTs produced in 1913 were capable of achieving resolutions of 0.01 °C or better up to 550 °C. Thermometers designed for geological and meteorological applications would have had reduced precision, naturally, but resolutions of 0.1 °C would have been easy to achieve.
The problem in 1913 wasn’t the precision or range of the thermometers. The main problem was that there wasn’t a nationally agreed upon temperature standard & scale. (The first ITS wasn’t published until 1927.) So while the precision of the various thermometers may have been high, their absolute accuracies would be in question. In other words, the data for the measured *change *in temperature for a given thermometer would have likely been valid, but you would you would not have high confidence in the results when comparing data from two different thermometers (unless they were “calibrated” by the same lab & person), and you would have even less confidence from the results of performing a direct comparison between temperature data generated in 1913 and a thermometer calibrated after 1927.
Are you sure about that? The NBS was established in 1901 and one of the early parameters that received a lot of attention was thermometry. I would have to think by 1913 any thermometer used for serious scientific purposes would have its calibration traceable to the NBS.
Research in thermometry had been occurring since the late 1600s. Very precise gas thermometers, LiG thermometers, and PRTs were made in the late 1800s. The question isn’t the precision of the thermometers, but their absolute accuracy. Without the adoption and adherence to an agreed-upon temperature standard & scale (which defines fixed point standards, traceability, methods for producing practical interpolation apparatuses, measurement uncertainties, etc.), you cannot compare temperature readings between various thermometers unless you also have a boatload of data that describes *exactly *how the thermometers were calibrated and certified.
So to answer your question… no, I am not 100% sure. But I haven’t seen any papers that describe how the accuracy and traceability of temperature measurements were assured during that time. Would love to know, as I think it’s an interesting subject.