Measuring Temperature by Sander Schimmelpenninck
sander@idirect.com

This is the first of several articles on how to observe weather phenomena, aimed at amateur meteorologists. I will draw on several weather books, advice from many experts in Canada’s Atmospheric Environment Service (AES) and elsewhere, and my background as an amateur since 1943. In that year somebody gave me a combination thermometer/hygrometer and a 1900 German book titled Meteorologie.

Let’s start with temperature, because it is the weather parameter most widely measured by even casual observers all over the world.

A little science

Temperature indicates the movement of molecules. That is pretty hard to do directly. Besides, who would want to?

From this it follows that you have a temperature of zero when the molecules in a substance are standing still. Scientists call that absolute zero, or 0 degrees on the Kelvin scale, named after an English physicist. It’s so cold that nature never comes close to it on Earth. You need very fancy lab equipment to approximate it.

For that reason man has developed several temperature scales that come closer to the human experience. The main ones in use today are Celsius (C) and Fahrenheit (F). On the Celsius scale, used in most countries except the United States, 0 equals the temperature of melting pure ice and 100 that of boiling water at sea level. It’s nice and logical, and it lets you test a thermometer in a simple home experiment. Celsius is also part of the metric system, which Canada adopted in the mid-1970s.

On the Fahrenheit scale, 32 is the temperature of melting ice and 220 that of boiling water at sea level pressure. I mention the latter because water boils at lower temperatures than 220 Fahrenheit or 100 Celsius at higher elevations. (Don’t open a thermos bottle in an unpressurized airplane at 10,000 feet. The coffee can boil over and scald you.)

Liquid in glass

Instruments that measure temperature come in a two basic categories: liquid in glass and solid-state

Liquid in glass refers to the thermometers with which we have all grown up. They include the instrument inserted into your mouth (and, in the old days, elsewhere) to measure fever. Their principle is simple. Liquids and solids react to temperature in highly-predictable fashion, according to their expansion coefficients. A bulb at the bottom of the thermometer contains a liquid, usually mercury or ethyl alcohol. Heat causes the liquid in the bulb to expand into a tube, made very thin to exaggerate expansion. The tube or its background is etched with one of more temperature scales. Mercury responds faster to temperature changes than ethanol does, but it freezes at -38 C, and costs more. Therefore many thermometers use ethanol, coloured for easy reading.

Meteorologists use three kinds of thermometers: normal, minimum, and maximum. A normal thermometer does what the name suggests: it shows the current temperature. (Well, it lags a bit, but most instruments do.)

A maximum thermometer uses mercury and has a constriction just above the bulb. The mercury rises along the scale until the temperature starts falling again. At that point the constriction causes the mercury column to break. The mercury below it shrinks back into the bulb, but the top portion stays put, which lets you see how warm things got. After reading, the observer or nurse swings the instrument, bulb down, to reunite the mercury in a manouevre called resetting.

A minimum thermometer uses clear or colored ethanol, with a sliding, dark-colored bar that fits inside the column. As the column contracts, its meniscus pulls the bar down with it until the temperature rises again. At that point the bar remains in place, so you can see how cold it got. After measuring the minimum temperature you tilt the instrument bulb-up so that the bar slides back to the meniscus.

Solid state

Today we associate solid state mostly with semiconductors and computer chips, but one form of solid-state temperature sensors existed long before transistors came along. They had a coil with one kind of metal on its outside surface and another, different one on the inside. Due to their different expansion coefficients, the coil tightens in low temperatures and unwinds when it’s warm. Attach a pointer to the coil, etch a scale along its path, and you have a bimetal temperature gauge. Many thermostats use the same principle.

Today’s solid-state temperature sensors have no moving parts. One basic type, the thermistor, changes its electrical resistance with temperature. That principle underlies the cheap but fairly accurate Radio Shack remote-reading temperature sensor. My neighbour has one in his front hallway, wired to a sensor under the eavestrough that overhangs his stoop. Fancier models also have a tiny computer that remembers temperature extremes since the last reset. The electrical characteristics of solid-state temperature sensors have made them the instruments of choice in the unmanned, automatic weather stations increasingly used by government and private meteorologists.

Close enough?

Two characteristics pertain to all measuring instruments: precision and accuracy. Many people confuse the two, so let’s define them here. Precision, also called resolution, is how finely you can read a phenomenon. The world’s most exquisite, diamond-studded wristwatch may have only hour and minute hands, giving a precision of one minute. For $30 you can buy a digital cheapie that reads ("resolves") to 1/00 of a second. So much for precision.

Accuracy refers to truth. Whether you measure in large steps or small ones, you want to know if you can believe your instrument. My $10, 20-year-old Temprite outdoor thermometer by Taylor reads to 0.5 Celsius. That also happens to be its accuracy: great value for money. At the other extreme, the $100 normal thermometers AES uses can be read to 0.05 C. However, they can err by 0.2 C at some points of the scale, so you need a correction chart to get the real temperature.

Installation

Shortwave radio enthusiasts know that a $200 Radio Shack shortwave set can outdo a $5,000 communications receiver, given the right antenna. Something similar applies in thermometry. Your thermometer or other sensor feels both the temperature of the surrounding ("ambient) air and radiated heat. The thin layer of air touching the bulb must have the same temperature as the air mass in the general vicinity, e.g. the air within one or several metres. That calls for good ventilation, natural or forced. In addition, direct radiation from the sun or a light bulb causes radiation to warm your sensor and thus makes it read too high.

To minimize these risks, meteorologists have long used the double-louvered, white shelters that resemble square beehives, called Stevenson screens after their British inventor. Some are ventilated by electric fans that suck in fresh air, to further reduce radiation and the formation of warm air pockets.

Few amateurs need or want a Stevenson screen. For everyday purposes, I recommend a pair of household outdoor thermometers, each mounted where you can expect shadow much of the day and good ventilation. The Temprite can be adjusted 1 degree C up or down for extra accuracy, but for that you need a high-quality reference thermometer.

This article is Copyrighted 1999: Sander Schimmelpenninck and used with his expressed written permission.  No unauthorized use allowed without prior written consent by him.