Mechanical vs Membrane Keyboards: What Is Actually Known
We could not find solid evidence that mechanical keyboards are faster or more accurate than membrane keyboards, or the reverse. Many web pages claim otherwise and quote studies, but the ones we tried to trace did not lead to a paper we could open, so we do not repeat them. What laboratory studies have measured, in controlled experiments, is something narrower and more useful: how much force a key needs to register a press. Keys needing about 1 newton produced noticeably more finger force and muscle activity than keys needing under half a newton. So choose a keyboard by its actuation force, feel and comfort, not by the word "mechanical".
What the difference is
In a mechanical keyboard every key has its own switch under it. In a membrane keyboard the keys press down on a rubber or silicone layer over a set of contacts. That describes the mechanism. It does not tell you how a keyboard feels: mechanical switches come with very different force, travel and sound, and membrane boards vary too. This is why the category label predicts little.
What the research actually tested
Two controlled laboratory studies are worth knowing. Both varied the force a key needs to register a press (the "make force", also called actuation force), and both measured muscle activity in the fingers with electromyography.
- Rempel and colleagues (1997). Ten experienced typists typed on three keyboards with make forces of 0.34, 0.47 and 1.02 newtons. There were no differences between the 0.34 and 0.47 N keyboards. Moving from 0.47 to 1.02 N raised applied fingertip force by about 40% and finger flexor muscle activity by about 20%. The authors suggested that keyswitches of 0.47 N or less should be considered over ones of 1.02 N, to keep the load on forearm tendons and muscles down. See the abstract.
- Gerard and colleagues (1999). Twenty-four female transcriptionists used their own keyboard (an audible click, 0.72 N) and three identical-design keyboards with no click and make forces of 0.28, 0.56 and 0.83 N, for 15 minutes each and then for seven workdays. Typing force and muscle activity were highest on the 0.83 N keyboard, and discomfort there was about 36% higher at the fingers, 40% higher in the lower arm and 39% higher overall. The lowest muscle activity was on the 0.28 N and the 0.72 N clicking keyboards. Seventeen of the 24 preferred the 0.72 N keyboard, four the 0.28 N one and three the 0.56 N one. See the abstract.
Read these results carefully. The samples were small, the tasks were short laboratory sessions, the keyboards were designs from the 1990s, and the outcome was muscle effort and comfort, not typing speed. Neither study compared "mechanical" with "membrane" as categories. What they support is modest: very stiff keys cost your fingers more, and preferences differ from person to person.
What that means for choosing a keyboard
Switch and keyboard specifications are often given in grams-force (gf) or centinewtons rather than newtons. Our own conversions of the figures above:
| Make force | Roughly in gf | Where it appeared |
|---|---|---|
| 0.28 N | 29 | Gerard: among the lowest muscle activity |
| 0.34 N | 35 | Rempel: no difference from 0.47 N |
| 0.47 N | 48 | Rempel: the suggested upper limit |
| 0.72 N | 73 | Gerard: lowest EMG (with audible click) and the most preferred |
| 0.83 N | 85 | Gerard: highest force, EMG and discomfort |
| 1.02 N | 104 | Rempel: about 40% more fingertip force than 0.47 N |
Two readings of that table are fair. Very stiff keys, from roughly 85 to 105 gf, were the costly ones. And lighter was not always better in people's own judgement, since most transcriptionists preferred the middle-high 0.72 N board with an audible click. So aim for moderate force and a feel you like, and check the number on the spec sheet when it is published.
One hardware detail that may matter for fast typists
In the typing study Observations on Typing from 136 Million Keystrokes (Dhakal, Feit, Kristensson and Oulasvirta, CHI 2018), most fast typists pressed the next key before releasing the previous one for 40 to 70% of their keystrokes. The paper notes that a keyboard that correctly senses any number of simultaneous presses is said to have "n-key rollover". It is our inference, not something the paper tested, that a keyboard which drops or misreads overlapping presses could hurt a fast typist. If you type quickly, check the rollover specification. This applies to both mechanical and membrane boards.
How to choose, practically
These are our suggestions:
- Try before you commit. If you can, type on it in a shop, or buy from a seller with an easy return policy. Feel is personal.
- Look at actuation force, travel and noise, not just the type. Noise matters in shared spaces.
- Consider your budget and layout, and whether you need a numeric keypad.
- Give yourself time to adjust. A new keyboard usually feels slower at first.
- Test it fairly. Take three runs of the same length and text type on the FreeTyper speed test on each keyboard, on different days, and compare the middle result and your accuracy, not a single run.
Setup matters more than the keyboard type
OSHA's keyboard guidance focuses on height, distance and wrist position, not on switch type: elbows near keyboard height, wrists straight, keyboard directly in front of you. A well-placed ordinary keyboard beats a poorly placed premium one. Our posture and wrist pain guide covers the setup.
What to be sceptical of
- "Mechanical keyboards add X WPM." We found no study we could open that shows this.
- Studies with oddly specific numbers and no citation. If a page does not name the journal, authors and year, treat it as unverified.
- A new keyboard as a fix for slow typing. Technique and accuracy matter far more. See how to type faster and how to improve typing accuracy.
Sources
- Rempel D, Serina E, Klinenberg E, Martin BJ, Armstrong TJ, Foulke JA, et al. The effect of keyboard keyswitch make force on applied force and finger flexor muscle activity. 1997.
- Gerard MJ, Armstrong TJ, Franzblau A, Martin BJ, Rempel DM. The effects of keyswitch stiffness on typing force, finger electromyography, and subjective discomfort. 1999.
- Dhakal V, Feit AM, Kristensson PO, Oulasvirta A. Observations on Typing from 136 Million Keystrokes. CHI 2018.
- OSHA, Computer Workstations eTool: Keyboards.
- The conversions to grams-force, the rollover inference and the buying checklist are our own.
Written by Ashiqur Rahman. Figures attributed to a source were checked against that source. If you spot something wrong, tell me and I will correct it.