Choosing and using thermometers

In the traditional 'wet' darkroom, temperature control is very important. At the very least you need repeatability; ideally, you also need accuracy as well. Then you need durability; an affordable price; and legibility.

6 thermometers

We strongly recommend a 'master/slave' approach, with at least two thermometers. The master is the most accurate and repeatable thermometer you can afford. The 'slave' is a cheaper thermometer, repeatedly checked against the 'master'.

In fact, we use two 'masters', a pair of mercury-in-glass Brannan laboratory thermometers accurate to 0.1 degrees C by interpolation, or 0.2 degrees on the scale. This enables us to check one against the other; they disagree by about 0.1 degree.

Six thermometers

Look closely and you can read all six -- though the thin columns of the mercury-in-glass thermometers are harder to read, illustrating the importance of legibility. You can also see that the spacing of the gradations varies widely; some are hard to read to even 0.5C, while others can be read (by interpolation) to 0.1C.

The actual readings are 17.4C; 64.5F (18C); 17.8C; 17.9C; 17.5C; and 64F (17.8C), a range of 0.6C, 1.1F, and an average (simple mean) of 17.7 +/- 0.3C.

The unusual colour of the markings on the Brannan (orange shading to black) is because the black paint had come out of the markings, so we used an old trick. Wipe some paint onto the thermometer; leave it a minute or two; them wipe it off. It sticks in the engravings, which are recessed and rough, but wipes off the shiny glass.

Accuracy and repeatability

It may seem odd to distinguish between accuracy and repeatability, but they are very different. An accurate thermometer, A, with absolute repeatability, will always show (for example) 20C as 20C. Another thermometer, B, less accurate but equally repeatable, may always show 20C as 19C. Yet a third, C, again fully repeatable, may always show 20C as 20.5C.

Obviously, if you are measuring with A and developing at 20C, you use the indicated temperature. With B, you will get exactly the same temperature with a marked 19C; with C, you use a marked 20.5C.

A thermometer with poor repeatability may show 19C one time, and 21C another (or worse). If you use an indicated 20C, you will never know exactly what temperature you are using.

thermometer dial


The best readily available thermometers for both accuracy and repeatability are usually mercury-in-glass, followed by spirit-in-glass. This is why our masters are mercury-in-glass.

Dial (bimetal) thermometers may or may not be as good for accuracy and repeatability. In our experience, good ones are highly repeatable, and can be adusted for accuracy; but ageing or a knock may necessitate re-adjustment.

Digital thermometers vary wildly in both accuracy and repeatability. There is a high correlation between quality and price. Top-quality digital thermometers are very good (but we still prefer to check them against a master), while the worst we have used are guaranteed to no more than +/- 1.5C in repeatability and +/- 1.5C in accuracy. In other words, 20 might read as anything between 17C and 23C, and even if you know that the thermometer reads a degree low, on average, you do not really know where you are between 17.5C and 20.5C.

How much accuracy do you need?

The maximum precision that is normally required for colour processing is +/- 0.25C, +/-0.5F. For black and white this is often relaxed to +/- 0.5C, +/1F.

Using the master/slave approach, both of these should be easily achievable. With a single thermometer, unless it is both highly repeatable and highly accurate, you are often working at the limits of accuracy of the instrument -- though this may not always matter, as described below under 'Compensation via time'.

Trying to work to greater accuracy than +/-0.25C is probably a mug's game -- see the notes below about 'fly by' -- and from experience (our own accidents, plus other photographers' stories) quite wide departures from optimum temperature will still give very acceptable colour slides, never mind printable negatives.

Compensation via time

A lack of accuracy may not matter, as long as the thermometer is repeatable. We have already seen how a master/slave approach compensates via the reading on the thermometer. Another possibility is that the photographer who always uses the same thermometer will compensate via time. Thus, for example, he notices that his negatives are always a little 'thin', and increases his development time by one minute to compensate. The truth is that he is developing for a little longer, at a slightly lower temperature.

This would happen with thermometer C, above, which shows 20C as 20.5C (reading 0.5 degree high). If he uses the 20C reading, he is actually developing at 19.5C.

The owner of thermometer B, on the other hand, might knock half a minute, so that he is developing for less time, at a higher temperature. Because his thermometer reads 1 degree low, showing 20C as 19C, an indicated 20C would be closer to a true 21C.

Fragility and expense

Mercury-in-glass and spirit-in-glass thermometers are fragile, especially lab-quality mercury-in-glass, so dial or digital thermometers are obviously more desirable for everyday use -- but it is as well to check them against the master, because they don't clearly break or suffer from column fragmentation, so you never know when they are mis-reading.

Lab-quality mercury-in-glass are also expensive and hard to find (and quite hard to read), and are therefore best kept as masters. For slaves, you can use lower-quality mercury-in-glass; spirit-in-glass; dial; or digital.

We use all of these. When we see what looks like a decent thermometer at a photo show, or otherwise second-hand, we buy it as a spare slave. We prefer spirit-in-glass or dial, simply because batteries in digital thermometers eventually die, usually at the start of a long week-end when you can't get another battery and haven't bought the spare that you've been meaning to buy for the last twelve months.

A great advantage of the master/slave relationship is that if you break the slave, you can use the master until you replace it (or switch to another slave). If you have only one thermometer, and break it, you can't do anything until you've bought another. Then, you don't know how closely your new thermometer is to the old one.

Speed of response

No immersion thermometer can give an instantaneous reading: inevitably, the thermometer itself has a thermal mass and must be heated to the temperature of the solution in which it is immersed. Response speeds vary widely, so it is worth taking a reading across (say) a couple of minutes, examining the thermometer every few seconds, to see how your thermometer responds. Typically the reading will shoot up (assuming the solution is warmer than the room) and then creep up for the last degree or so. When it starts falling again, you can assume that the solution is cooling.

thermometer  glass and cork

Paterson spirit thermometer

A good middle-of-the-road thermometer, easy to read, marked in both F and C. The cork on the top is a useful 'buffer' and handle, and stops the thermometer disappearing into a Nova slot processor. This is one of our best 'slaves', reading 0.5C below our masters, i.e. a marked 19.5C is a true 20C. The 'Certified' version should be more accurate but is no more consistent.

Consistency in use

Lab thermometers normally specify an immersion depth, typically 10cm/4 inches. Obviously this is not super-critical, but equally, for maximum repeatability, you want a more or less constant immersion depth: just dipping the bulb under the surface will give a different reading from submerging it 10cm, especially if there is a marked difference between air temperature and solution temperature (as when processing E6 in a cool room). Likewise, the water bath in our Jobo CPE-2 (which has no water circulation pump) will give different readings in different places. These are stable, so one reading from one place is all you need. But it should always be the same place.

Column fragmentation

Mercury-in-glass and spirit-in-glass thermometers can suffer from column fragmentation -- little gaps between the liquid in the column. With mercury you may be able to shake the liquid down, but we have an aversion to swinging fragile mercury-in-glass thermometers around, even out of doors.

A better approach is a beaker of very hot (but not boiling) water, around 40-50C, 110-130F. Dip the bulb of the thermometer into it repeatedly, until the liquid begins to reach the small bulb at the top. You can see this small secondary bubble very clearly in the second thermometer from the left in the first picture.

VERY CAREFULLY repeat the dipping until the last gap has gone up into the upper bulb. It is quite possible to break the thermometer this way, especially spirit thermometers where the spirit has a very low boiling point. But if you do it carefully, the thermometer can be made as good as new.

Do not do what a novice friend of ours did, and assume that the top of the fragmented column is what you read. When we cured her thermometer, as described above, it was nearly 5 degrees lower, i.e. she had been developing at 15C instead of 20C.

Thermal gradients

The greater the thermal gradient (the difference between the solution temperature and the room temperature), the faster the solution will cool (or, more rarely, heat up). It is therefore a good idea, wherever possible, to work as close to room temperature as possible (e.g. 20C/68F in a cool room, 24C/75F in a hot room) or to use a water bath, or both.

development tanks

Development tanks and 'fly by'

If you tip warm developer into a cold development tank, obviously the temperature of the developer drops as it heats the tank, spools and film. The greater the thermal mass of the tank, spools and film, the more this matters. This is where stainless steel has a marked advantage: there is less mass, and it hold the heat less, so it comes to temperature more quickly.

Kindermann and Paterson tanks

Our stainless tanks (Kindermann) and reels (Hewes) warm and cool much faster than plastic (Paterson and Jobo).

Some use pre-soaks to bring the tanks to temperature. We don't. This is because many films have chemicals in the emulsion that are designed to facilitate development, and these may be washed out in a pre-soak. On the other hand, while most manufacturers give the advice not to pre-soak, they freely admit that the advice is based more on theory than on practice, and that with most films, pre-soaks do little or no harm.

What we do instead is to take the temperature of the developer both before and after development; the difference can be impressive. Typically, if developing at 24C with our CPE-2, we'll put the dev in at 24.5C and it will come out at 23.5C: an average of 24C.

Non-contact thermometers

You can buy non-contact thermometers that are alleged to give accurate temperature readings when held some distance above the solution. We have never used these, as they seem to be an excessively complicated solution with large possibilities for error, but some people use them and swear by them. We mention them only for completeness.

Mercury and health

Mercury is moderately unpleasant stuff, but equally, it is possible to become overly excited about it. The risk of exposure is very small, unless you break the thermometer, and even then, you're looking at exposure to small quantities for a short time. Probably the best way to dispose of mercury waste from a broken thermometer is to put it in a small bottle (scoop it up on as piece of paper, using another piece of paper as a pusher) and take it to your local dispensing chemist/pharmacist/drugstore.

The bottom line

The fact that it is possible to write an entire short module on thermometers is a clear warning that one could easily become obsessive about this aspect of photography -- as indeed one can about many others. We hope that the above gives useful guidance about how to achieve as much consistency as you need, as conveniently and economically as possible.

Go to the list of modules

or go to the home page

or support the site with a small donation.


© 2010 Roger W. Hicks