Should we trust an experiment that suggests a stone can eat iron?

Is it poor scientific practice to explain away results we would not expect?

Keith S. Taber

how convinced would be be by a student who found an increase in mass after burning some magnesium and argued that this showed that combustion was a process of a substance consuming oxygen as a kind of food

I came across an interesting account of an experiment which seemed to support a hypothesis, but where the results were then explained away to reject the hypothesis.

Experimental results always need interpretation

That might seem somewhat dubious scientific practice, but one of the things that becomes clear when science is studied in any depth is that individual experiments seldom directly decide scientific questions. Not only is the common notion that a positive result proves a hypothesis correct over-simplistic, but it is also seldom the case that a single negative result can be assumed to be sufficient to reject a hypothesis. ¹

Given that, the reason I thought this report was interesting is that it was published some time ago, indeed in 1600. It also put me in mind of a practical commonly undertaken in school science to demonstrate that combustion involves a substance combining with oxygen. In that practical activity (commonly mislabelled as an 'experiment' ²), magnesium metal (for example) is heated inside a ceramic crucible until it has reacted, and by careful weighing it is found (or perhaps I should say, it should be found, as it can be a challenging practical for the inexperienced) that the material after combustion weighs more than before – as the magnesium has reacted with a substance from the air (oxygen).³ This is said to give support to the oxygen theory of combustion, and to be contrary to the earlier phlogiston theory which considered flammable materials to contain a substance called phlogiston which was released during combustion (such that what remains is of less mass than before).

Testing whether lodestones eat iron

The historical experiment that put me in mind of this involved burying a type of stone known as a lodestone in iron filings. The stone and filings were carefully weighed before burial and then again some months later after being separated. The hypothesis being tested was that the weight of the lodestone would increase, and there would be a corresponding decrease in the mass of the weight of the iron filings. Apparently at the end of the experiment the measurements, strictly at least, suggested that this was what had occurred. Yet, despite this, the author presenting the account dismissed the result – arguing that it was more likely the finding was an artifact of the experimental procedure either not being sensitive enough, or not having been carried out carefully enough.

Explaining away results – in science and in school laboratories

That might seem somewhat against the spirit of science – I wonder if readers of this posting feel that is a valid move to make: to dismiss the results, as if scientists should be fee to pick and chose which results they wish to to take notice of?

But I imagine the parallel situation has occurred any number of times in science classrooms, for example where the teacher responds to students' practical demonstrations that what is left after burning magnesium has less mass than the magnesium had before. Rather than seeing this as a refutation of the oxygen hypothesis (actually now, of course, canonical theory) – and possible support for the notion that phlogiston had been released – the teacher likely explains this away as either a measurement error or, more likely, a failure to retain all of the magnesia [magnesium oxide] in the crucible for the 'after' measurement.

Hungry magnets

The historical example is discussed in William Gilbert's book about magnetism, usually known in English as 'On the magnet'. ⁴ This is sometimes considered the first science book, and consists of both a kind of 'literature review' of the topic, as well as a detailed report of a great many observations and demonstrations that (Gilbert claims) were original and made by Gilbert himself. There were no professional scientists in 1600, and Gilbert was a physician, a medical practitioner, but he produced a detailed and thoughtful account of his research into magnets and magnetism.

Gilbert's book is fascinating to a modern reader for its mixture of detailed accounts that stand today (and many of which the reader could quite easily repeat) alongside some quite bizarre ideas; and as an early example of science writing that mixes technical accounts with language that sometimes seems quite unscientific by today's norms – including (as well as a good deal of personification and anthropomorphism) some very unprofessional remarks about some other scholars he considers mistaken. Gilbert certainly has little time for philosophers ('philosophizers') who set out theories about natural phenomena without ever undertaking any observations or tests for themselves.

Lodestones

Magnetism has been known since antiquity. In particular, some samples of rock (usually samples of magnetite, now recognised as Fe₃O₄) were found to attract both each other and samples of iron, and could be used as a compass as they aligned, more or less, North-South when suspended, or when floated in water (in a makeshift 'boat'). Samples of this material, these naturally occurring magnets, were known as lodestones.

Yet the nature of magnetism, seemingly an occult power that allowed a stone to attract an iron nail, or the earth to turn a compass needle, without touching it, remained a mystery. Some of the ideas that had been suggested may seem a little odd today.

Keepers as nutrients?

One idea, that Gilbert dismisses is that the magnet (lodestone) in effect needs a food source to keep up its strength,

"The loadstone is laid up in iron filings, not that iron is its food; as though loadstone were alive and needed feeding, as Cardan philosophizes; nor yet that so it is delivered from the inclemency of the weather (for which cause it as well as iron is laid up in bran by Scaliger; mistakenly, however, for they are not preserved well in this way, and keep for years their own fixed forms): nor yet, since they remain perfect by the mutual action of their powders, do their extremities waste away, but are cherished & preserved, like by like."

Gilbert, 1600 – Book 1, Chapter 16.

Girolamo Cardano was an Italian who had written about the difference between amber (which can attract small objects due to static electrical charges) and lodestones, something that Gilbert built upon. However, Gilbert was happy to point out when he thought 'Cardan' was mistaken.

An experiment to see if iron filings will feed a magnet

Gilbert reports an experiment carried out by Giambattista della Porta. Porta's own account is that:

"Alexander Aphrodiseus in the beginning of his Problems, enquires wherefore the Loadstone onely draws Iron, and is fed or helped by the fillings of Iron; and the more it is fed, the better it will be: and therefore it is confirmed by Iron. But when I would try that, I took a Loadstone of a certain weight, and I buried it in a heap of Iron-filings, that I knew what they weighed; and when I had left it there many months, I found my stone to be heavier, and the Iron-filings lighter: but the difference was so small, that in one pound I could finde no sensible declination; the stone being great, and the filings many: so that I am doubtful of the truth."

Porta, 1658: Book 7, Chapter 50

Gilbert reports Porta's experiment in his own treatise, but adds potential explanations of why the iron filings had slightly lost weight (it is very easy to lose some of the material during handling), and why the magnet might be slightly heavier (it could have become coated in some material during its time buried),

"Whatever things, whether animals or plants, are endowed with life need some sort of nourishment, by which their strength not only persists but grows firmer and more vigorous. But iron is not, as it seemed to Cardan and to Alexander Aphrodiseus, attracted by the loadstone in order that it may feed on shreds of it, nor does the loadstone take up vigour from iron filings as if by a repast on victuals [i.e., a meal of food]. Since Porta had doubts on this and resolved to test it, he took a loadstone of ascertained weight, and buried it in iron filings of not unknown weight; and when he had left it there for many months, he found the stone of greater weight, the filings of less. But the difference was so slender that he was even then doubtful as to the truth. What was done by him does not convict the stone of voracity [greediness, great hunger], nor does it show any nutrition; for minute portions of the filings are easily scattered in handling. So also a very fine dust is insensibly born on a loadstone in some very slight quantity, by which something might have been added to the weight of the loadstone but which is only a surface accretion and might even be wiped off with no great difficulty."

Gilbert, 1600 – Book 2, Chapter 25.

Animistic thinking

To a modern reader, the idea that a lodestone might keep up its strength by eating iron filings seems very fanciful – and hardly scientific. To refer to the stone feeding, taking food, or being hungry, is animistic – treating the stone as though it is a living creature. We might wonder if this language is just being used metaphorically, as it seems unlikely that intelligent scholars of the 16th Century could actually suspect a stone might be alive. Yet, as Gilbert points out, there was a long tradition of considering that the lodestone, being able to bring about movement, had a soul, and Gilbert himself seemed to feel this was not so 'absurd'.

A reasonable interpretation?

We should always be aware of the magnitude of likely errors in our measurements, and not too easily accept results at the margins of what can be measured. Gilbert's suggestions for why the test of whether mass would be transferred from the iron to the magnet might have given flawed positive results seem convincing. It would be easy to lose some of the filings in the experiment: especially if the "heap of Iron-filings" was left for several months without any containment! And the lodestone could indeed easily acquire some extraneous material that needed to be cleaned off to ensure a valid weighing. As the lodestone attracts iron, all of the filings would need to be carefully cleaned from it (and returned to the 'heap' before the re-weighing).

But, I could not help but wonder if, in part at least, I found Gilbert's explaining away of the results as reasonable, simply because I found the premise of the iron acting as a kind of food as ridiculous. We should bear in mind that although the predicted change in mass was motivated by a notion of the magnet needing nutrition, that might not be the only scenario which might give rise to the same prediction. ¹ After all, how convinced would be be by a student who

• suggested combustion was a process of a substance consuming oxygen as a kind of food, and
• therefore predicted that magnesium would be found to have got heavier after a good meal, and
• subsequently found an increase in mass after burning some magnesium, and
• argued that this gave strong support for the oxygen-as-food principle?

Coda

It is rather difficult for us today to really judge how language was used centuries ago. Do these natural philosophers talking of magnets eating iron mean this literally, or is it just figurative – intended as a metaphor that readers would understand suggested that there was a process somewhat akin to when a living being eats? ⁵ Some of them seemed quite serious about assigning souls to entities we today would conspire obviously inanimate. But we should be careful of assuming apparently incredible language was meant, or understood, literally.

In the same week as I was drafting this posting I read an article in Chemistry World about how the heavier elements are produced, which quoted Professor Brian Metzer, physicist at Columbia University,

"What makes the gamma-ray burst in both of these cases [merging neutron stars and the collapse of large rapidly rotating stars] is feeding a newly-formed black hole matter at an extremely high rate…The process that gives rise to the production of this neutron-rich material is actually outflows from the disc that's feeding the black hole."

Brian Metzer quoted in Wogan, 2022

If we would be confident that Professor Metzer meant 'feeding a black hole' to be understood figuratively, we should be careful to reserve judgement on how the feeding of lodestones was understood when Porta and Gilbert were writing.

Sources cited:

• Gilbert, W. (1600/1900). On the Magnet, Magnetick Bodies also, and on the great magnet of the earth; a new physiology, demonstrated by many arguments & experiments (S. P. Thompson, Trans. Project Gutenberg ed.). [Available as a free e-book at https://www.gutenberg.org/files/33810/33810-h/33810-h.htm]
• Porta, G. d. (1658). Natural magick by John Baptista Porta, a Neapolitane ; in twenty books … wherein are set forth all the riches and delights of the natural sciences. On-line version at https://quod.lib.umich.edu/e/eebo/A55484.0001.001/1:10?rgn=div1;view=toc
• Taber, K. S. and Watts, M. (1996) The secret life of the chemical bond: students' anthropomorphic and animistic references to bonding, International Journal of Science Education, 18 (5), pp.557-568.
• Wogan, T. (2022, 5th Janaury 2022). How elements are made beyond the stars. Chemistry World, 19, 28-31.

Notes:

¹ Strictly scientific tests never 'prove' or 'disprove' anything.

The notion of 'proof' is fine in the context of purely theoretical disciplines such as in mathematics or logic, but not in science which tests ideas empirically. Experimental results always underdetermine theories (that is, it is always possible to think up other theories which also fit the results, so a result never 'proves' anything). Apparently negative results do not refute ('disprove') a theory either, as any experimental test of a hypothesis also depends upon other factors (Has the researcher been sloppy? Is the measuring instrument valid – and correctly calibrated? Are any simplifying assumptions reasonable in the context…). So experimental results offer support for, or bring into question, specific theoretical ideas, without ever being definitive.

² An experiment is undertaken to test a hypothesis. Commonly in school practical work 'experiments' are carried out to demonstrate an accepted principle, such that it is already determined what the outcome 'should' be – students may have already been told the expected outcome, it appears i n their textbooks, and the title of the activity may be suggestive ('to show that mass increases on combustion'). Only if there is a genuine uncertainty about the outcome should the activity be labelled an experiment – e.g., it has been suggested that combustion is like the fuel eating oxygen, in which case things should be heavier after burning – so let's weigh some magnesium, burn it, and then re-weight what we have left (dephlogisticated metal?; compound of metal with oxygen?; well-fed metal?)

³ Mass and weight are not the same thing. However, in practice, measurements of weight made in the laboratory can be assumed as proxy measurements for mass.

⁴ As was the norm in European scholarship at that time, Gilbert wrote his treatise in Latin – allowing scholars in different countries to read and understand each other's work. The quotations given here are from the 1900 translation into English by S.P. Thompson.

⁵ Such metaphors can act as communication tools in 'making the unfamiliar familiar' and as thinking tools to help someone pose questions (hypotheses?) for enquiry. There is always a danger, however, that once such figures of speech are introduced they can channel thinking, and by providing a way of talking about and thinking about some phenomena they can act as obstacles to delving deeper in their nature (Taber & Watts, 1996).