Soddy: Primer

What this project is. Frederick Soddy (1877-1956; Nobel Prize in Chemistry 1921, awarded 1922 [memory, high]) wrote the two popular books in the folder: Matter and Energy (1911) and The Interpretation of Radium and the Structure of the Atom (4th edition, August 1920; 1922 reprint). With Rutherford at McGill in 1902-03 he worked out that radioactive atoms change into other atoms. He later named and explained isotopes. These are not fringe books. They are the founding popular account of a theory that turned out to be right, written by one of its authors. They also contain a few instructive wrong turns: a heating Earth that would one day explode, gold and silver in one decay chain, a forgotten race that mastered transmutation, and an estimate that atomic energy would need millions of volts. The primer sets out the physics that held up, shows where the inferences went further than the evidence, and uses Soddy's own numbers for the worked example.

Source gap (carried from claims.md and the arbiter). The folder holds only the two popular books. Soddy's research papers (with Rutherford, with Ramsay, on lead and on isotopes) and his economics books (Cartesian Economics 1922, Wealth, Virtual Wealth and Debt 1926 and later) are not in the folder. The Investigator found some of them online and read extracts only. Whatever this primer says about them rests on those extracts [web] or on [memory], and it says so each time. The deep reads cover Interpretation PDF pp.49-57, 106-132, 141-146, 153-165, 169, 195, 207-221 and 268-300, and Matter and Energy PDF pp.13-42, 101-118 and 201-257. The arbiter read Interpretation PDF pp.247-251 (actinium) itself and checked passages against the 1909 first edition on archive.org. The rest of both books rests on page-by-page digests, plus passages the reviewers read themselves. In particular, Interpretation PDF pp.252-256 (protactinium, the complete series tables) is still digest-only [arbiter section 10.4].

Tags. [read] = from the books (PDF page numbers, as extracted; "printed p." where the book's own page number matters); [web] = from an Internet source, URL given; [memory] = from model training, with confidence; [calc] = arithmetic redone for this primer (see example.py). The verdicts are the arbiter's (arbiter.md). This primer adds no new verdicts.

Reader's key to Soddy's names. Soddy used the names of his day. In modern terms:

Warning about Soddy's "periods". Soddy's "periods" are mean lives, not half-lives, so they look 1.44 times longer than the half-lives quoted in modern tables. Radium's 2,500 years is a mean life; his half-life for it is about 1,700 years [read: Interp. PDF p.171].

---

1. The idea in one paragraph

Radium keeps giving off heat, a gram of it enough to bring its own weight of water from freezing to boiling in under an hour, year after year, with no fuel. Soddy and Rutherford's answer was that the atoms themselves are changing. Each radium atom has a fixed small chance every second of blowing apart. It throws out a helium nucleus at high speed and becomes a different element, a heavy gas. That gas breaks down in turn, through a chain of short-lived elements, and the chain ends in lead. The heat is the energy of those explosions, so it is not taken from anywhere outside the radium. Per gram it is nearly a million times more energy than any chemical reaction gives. Radium has not all vanished because uranium keeps making it. A slow parent and a fast daughter settle into a balance, like a tank filled by a slow tap and emptied by a fast drain. Following the chain led Soddy to two more ideas. Atoms of one chemical element can have different weights (he named them "isotopes"). And if people ever learned to start or speed up these changes, the energy locked in atoms would transform civilisation, for good or for very great harm.

2. What you need to know first

First-order differential equations and exponential decay. If something disappears at a rate proportional to how much of it is left, dN/dt = -lambda N, then N(t) = N_0 e^(-lambda t). Here N is the number of atoms left, t is time, N_0 is the starting number and lambda (Greek "lambda") is a constant with units of 1/time. This is the same equation that describes a capacitor discharging through a resistor. Read: any calculus or ODE text, the section on separable equations; Boas, Mathematical Methods in the Physical Sciences, the ordinary differential equations chapter.

Probability and averages over a continuous distribution. If the chance that an atom decays between t and t + dt is lambda e^(-lambda t) dt, the average lifetime is the integral of t times that, which comes out to 1/lambda. You need to be comfortable with that kind of integral (integration by parts). Read: Taylor, An Introduction to Error Analysis, the chapters on distributions; or any probability text on the exponential distribution.

Atoms, moles and the ideal gas. One mole is 6.022 x 10^23 atoms (Avogadro's number, N_A). A mole of any ideal gas at 0 C and 1 atmosphere ("STP", standard temperature and pressure) fills 22.4 litres = 22,400 cm^3. One gram of radium-226 holds N_A / 226 = 2.66 x 10^21 atoms. Read: any first-year chemistry text, the chapter on gases; Schroeder, An Introduction to Thermal Physics, chapter 1.

Energy units: calories, joules, electron volts. A calorie (cal) is the heat that warms 1 g of water by 1 C, 4.184 joules (J). An electron volt (eV) is the energy a charge e (1.602 x 10^-19 coulomb) gains when it crosses 1 volt: 1 eV = 1.602 x 10^-19 J; 1 MeV = 10^6 eV. Chemical reactions release a few eV per atom. Nuclear decays release a few MeV, about a million times more. A particle of charge q accelerated through V volts gains energy qV. An alpha particle has charge 2e, so 2 million volts gives it 4 MeV.

Mass-energy, E = mc^2. When a nucleus decays, the products weigh slightly less than the parent. The missing mass Delta m appears as energy Q = Delta m c^2. Q is called the "Q-value" of the decay, and c = 3.00 x 10^8 m/s is the speed of light. One atomic mass unit (u) is equivalent to 931.5 MeV. Einstein published the relation in 1905 [memory, high], but in 1911-1920 nobody knew how to calculate the energy of a nucleus from it. Read: Serway, Moses and Moyer, Modern Physics, the special-relativity and nuclear-structure chapters; or Krane, Introductory Nuclear Physics, the chapter on nuclear masses and binding energy.

Atomic number and mass number. Every nucleus has Z protons (the atomic number, which fixes the chemistry) and A nucleons (the mass number: protons plus neutrons). The neutron was discovered only in 1932 [memory, high], so Soddy's books speak of "atomic weight" and "net nuclear charge" instead. An alpha particle is a helium-4 nucleus (Z = 2, A = 4). A beta particle is an electron emitted by the nucleus. Read: Krane, the opening chapter on basic nuclear concepts.

Coulomb's law and the electrostatic barrier. Two positive charges Z_1 e and Z_2 e at distance r have potential energy U = Z_1 Z_2 e^2 / (4 pi epsilon_0 r), where epsilon_0 is the permittivity of free space. In nuclear units, e^2 / (4 pi epsilon_0) = 1.44 MeV x fm, where 1 fm (femtometre) = 10^-15 m. Read: Griffiths, Introduction to Electrodynamics, the electrostatics chapter.

Quantum tunnelling (only as an idea). In quantum mechanics a particle can cross an energy barrier higher than its own energy, with a probability that falls off exponentially with the barrier's height and width. You do not need the mathematics here. You need the consequence: nuclear reactions can happen below the "classical" barrier energy. Read: Griffiths, Introduction to Quantum Mechanics, the WKB-approximation chapter, which works Gamow's theory of alpha decay as an example.

3. The textbook view

3.1 The decay law

Each unstable nucleus has a fixed probability lambda per unit time of decaying. That probability does not depend on how long the nucleus has already existed. For N nuclei:

• dN/dt = -lambda N, so N(t) = N_0 e^(-lambda t)
• activity (decays per second) = lambda N. The unit is the becquerel (1 decay/s). The old unit, the curie (3.7 x 10^10 decays/s), was defined as the activity of one gram of radium [memory, high].
• mean life T = 1/lambda. Proof: the average of t over the probability density lambda e^(-lambda t) is the integral from 0 to infinity of t lambda e^(-lambda t) dt = 1/lambda.
• half-life t_1/2 = (ln 2)/lambda = 0.693 T. So T = 1.4427 t_1/2. Soddy prints this factor as "1.45" [read: Interp. PDF p.159].
• After one mean life, a fraction e^(-1) = 0.368 is left. Soddy prints this [read: Interp. PDF p.159].

"Independent of age" is a strong statement. It means the exponential distribution is memoryless: the chance of surviving another s seconds is e^(-lambda s), whatever the atom's age. Quantum mechanics explains why. An unstable nucleus is a single quantum state with a fixed decay amplitude. For alpha decay, Gamow (1928) showed that the alpha particle tunnels out through the Coulomb barrier [memory, high]. The rate is almost completely independent of chemistry, temperature and pressure. There is one known exception: decays that involve the atom's own electrons. In electron capture the nucleus absorbs one of its own electrons; beryllium-7 changes rate by about 1% between chemical compounds [memory, medium]. Bound-state beta decay is the extreme case. Fully stripped rhenium-187 has a half-life of about 33 years, against about 41 billion years for the neutral atom (Bosch et al. 1996) [web: https://www.osti.gov/biblio/404092].

3.2 Chains and equilibrium

A parent (decay constant lambda_1, N_1 atoms) produces a daughter (lambda_2, N_2 atoms):

dN_2/dt = lambda_1 N_1 - lambda_2 N_2

The first term is production and the second is loss. If the parent lives much longer than the daughter, N_1 is effectively constant. Then N_2 grows as N_2(t) = (lambda_1 N_1 / lambda_2)(1 - e^(-lambda_2 t)) and levels off at

lambda_1 N_1 = lambda_2 N_2, so N_2/N_1 = lambda_1/lambda_2 = T_2/T_1 (secular equilibrium).

In words: at equilibrium the parent and the daughter have the same activity, and their amounts are in the ratio of their mean lives. The balance carries down the whole chain (lambda_1 N_1 = lambda_2 N_2 = lambda_3 N_3 = ...), so the rule links the parent to any descendant. If the parent is only somewhat longer-lived, the ratio becomes N_2/N_1 = lambda_1/(lambda_2 - lambda_1) ("transient equilibrium"). Soddy states the secular case and its condition correctly but does not mention the transient case [read: Interp. PDF p.162; mathematician].

The rule works in both directions. If you know the daughter's life and can measure the ratio of amounts, you get the parent's life. This is how lives too long to watch were measured: radium's from its emanation, and uranium's from radium.

3.3 The displacement rules and isotopes

• Alpha decay: Z falls by 2 and A falls by 4. Example: Ra-226 (Z = 88) gives Rn-222 (Z = 86) and He-4.
• Beta-minus decay: Z rises by 1 and A stays the same. A neutron becomes a proton and emits an electron and an antineutrino. (An antineutrino is a neutral particle of almost zero mass that interacts so weakly with matter that it passes through the whole Earth. It carries off part of the decay energy, so that energy never shows up as heat. It was proposed by Pauli in 1930 and detected in 1956 [memory, high], so it is absent from Soddy's books.)
• One alpha plus two betas, in any order, returns the atom to the same Z, 4 mass units lighter. Its chemistry is unchanged, so it is an isotope of the parent.
• Decay never raises A. The uranium-238 chain to lead-206 takes 8 alphas (238 - 32 = 206) and 6 betas (92 - 16 + 6 = 82) [calc].

Chemistry depends only on Z, through the electrons. So isotopes are chemically identical and can be separated only by mass-dependent methods (diffusion, mass spectrometry, centrifuges).

An example from Soddy's own book: the actinium "branch". Actinium is always found in uranium ores, but its alpha activity there is "only about one-fifteenth or one-sixteenth" of the uranium series, where five-eighths would be expected if it sat in the main line [read: Interp. PDF pp.247-250, read by the arbiter; no deep-read file]. Soddy offered two explanations: actinium is "an entirely separate and independent primary radio-element" (but then why is it always with uranium?), or uranium's chain branches, "out of every twelve uranium atoms, eleven go through the main line ... and one goes through the actinium line" (about 8 per cent). The answer, found only in 1935 when Dempster discovered uranium-235 [web, summary level: https://en.wikipedia.org/wiki/Uranium-235], is that both alternatives are half right. Actinium's ancestor is a separate primary parent, and it is uranium: a second isotope (0.72% of natural uranium) that no chemical test could tell apart from uranium-238. With the modern figures the actinium series gives about 1/25 of the uranium series' alpha activity, so Soddy's observed 1/15-1/16 was high by a factor of about 1.6, and his 8 per cent "branch" by about 1.8 [calc, arbiter section 10.2]. The arbiter's verdict on the branch hypothesis is Historically superseded but instructive: chemistry alone cannot tell a branch from a hidden isotope, because the two behave identically in the test tube. (Check yourself, question 5, runs this calculation backwards.) In the same chapter Soddy reports the 1918-19 discovery of actinium's parent, now protactinium, by "Cranston and the writer" and by Hahn and Meitner [read: Interp. PDF p.252, digest only]. This is a second priority statement by Soddy, who had a stake in it. A modern summary says Hahn and Meitner's paper is dated 1918, that Soddy and Cranston had reported indirect evidence less than three months earlier, and that credit usually goes to Hahn and Meitner [web, summary level: https://en.wikipedia.org/wiki/Protactinium, via the arbiter]. The dated papers themselves were not read.

3.4 Where the energy comes from

The energy released in one decay is the Q-value, Q = (M_parent - M_products) c^2. In alpha decay most of Q goes to the alpha's kinetic energy and a little to the recoiling nucleus. In beta decay, Q is shared among the electron, an antineutrino and often gamma rays. The antineutrino leaves any calorimeter, so a heat measurement always captures a little less than the total Q.

Masses are not exact whole numbers. The small departures ("mass defect", or Aston's 1927 "packing fraction" [memory, high]) are the nuclear binding energy, and they are where the energy comes from.

3.5 Getting into a nucleus

To make a nucleus react, a charged projectile has to come within reach of the strong force, a few fm. The classical Coulomb barrier for an alpha particle on a target of charge Z and mass number A is about

V_C = 2 Z x 1.44 MeV fm / [1.2 fm x (A^(1/3) + 4^(1/3))]

For nitrogen-14 this gives about 4.2 MeV, and for copper-63 about 12.5 MeV [calc, arbiter]. Natural alpha particles carry 4 to 8 MeV, so they can hit light nuclei such as nitrogen but cannot reach heavy ones. Thanks to tunnelling, reactions happen with some probability below V_C. That is why Cockcroft and Walton could split lithium in 1932 with protons of only about 0.7 MeV [memory, high].

The bridge Soddy printed without knowing it: the Geiger-Nuttall rule. Tunnelling also works in the other direction, letting an alpha particle out of a nucleus. Soddy prints a table of alpha ranges in air (25 to 69.5 mm) and says that if you plot the logarithm of each emitter's period against the logarithm of its alpha range, you get three parallel straight lines, one for each series, "the reason ... still obscure" [read: Interp. PDF p.210, deep read]. This is the Geiger-Nuttall law (Geiger and Nuttall, 1911-12), and its reason was supplied in 1928 by Gamow and by Gurney and Condon: a faster alpha (longer range) meets a thinner part of the barrier, and the tunnelling probability depends on that thickness exponentially, so a small rise in energy shortens the life enormously [memory, high]. The line is very steep: the deep read's fit of Soddy's own table gives period proportional to roughly range to the power -52 [calc, deep read]. Soddy used the rule to estimate ionium's life at 200,000 years before it was measured at 100,000, and uranium II's at two to three million years, against a modern mean life of about 354,000 years [read: pp.195, 199, 208, 211; memory, high for the modern value]. On so steep a line, errors of a factor of three or more are the expected accuracy, and Soddy called the estimates "approximate". Arbiter verdict on the rule as Soddy states it: Established (mechanism explained 1928).

Ramsay and Cameron: a transmutation claim that was tested and failed. In 1907 William Ramsay (Nobel Prize in Chemistry 1904 [memory, high]) and Cameron reported that radium emanation turned copper into lithium. Curie and Gleditsch traced the lithium to the glass vessels, and Soddy reported his own negative copper and gold results to the Chemical Society in 1909 [web, snippets via the Investigator; the original papers were not read]. The barrier formula above explains why it could not work: a 5.5 MeV alpha particle from the emanation faces a 12.5 MeV barrier at copper, and getting from copper (Z = 29) to lithium (Z = 3) would mean removing 26 protons. Arbiter verdict: Tested, not supported; the proposed mechanism Refuted by calculation [arbiter section 10.2, C6e]. It was rejected by experiment, not by authority, and it is the backdrop to Rutherford's 1919 success with nitrogen.

3.6 Earth and sun (needed for claim 7)

The Earth loses about 44-47 terawatts (TW) through its surface [web: https://en.wikipedia.org/wiki/Earth%27s_internal_heat_budget]. Geoneutrino detectors (KamLAND, Borexino) measure about 20 TW from uranium and thorium, so roughly half the heat is radiogenic [web: Gando et al. 2011, via the contrarian]. Radioactive elements are concentrated in the crust and upper mantle, not spread evenly through the globe, and the mantle convects. So on the whole the Earth is cooling, not heating up. The sun radiates 3.8 x 10^26 W. Its power comes from hydrogen fusing to helium (Eddington's 1920 suggestion; Bethe's mechanism 1938-39 [memory, high]), not from radioactivity.

4. Where Soddy departs

Soddy mostly does not depart from the textbook. The textbook was built partly from his work. The departures are at the edges, where he took a correct principle one step past the evidence. The table puts each departure next to the modern view. The entries in the middle column are his own words, and the ones in the right column carry the arbiter's verdicts.

The gold-silver idea, worked through. Soddy's premise is the equilibrium rule from section 3.2: in one chain, amount is proportional to mean life. Gold is much rarer than silver, so by that rule gold would be the shorter-lived member, the daughter, and silver the parent. But silver (A = 107 or 109) would then have to decay into gold (A = 197). That means gaining 88 or 90 mass units, and decay never raises A. Swap the roles to make the chain possible (gold decaying down to silver), and gold becomes the longer-lived parent, so it should be the more abundant of the two. That contradicts the premise. Either way the idea fails under Soddy's own displacement rules [calc, physicist, arbiter]. To be fair to him: he writes "at least a possible view" and "Obviously we are only at the beginning", and at the time no stable isotope's mass was known. The passage carries a footnote on gold's unstable purchasing power that cites the monetary reformer Arthur Kitson [read: Interp. PDF p.203]. It is the hinge between Soddy's physics and his later economics. Separately, a 1905 letter shows Soddy asked to examine British Museum coins "for chemical reaction" and was refused (CSAC catalogue item 217) [web: research/2026-10-08-archiveshub-csac-catalogue.md; letter unread]. Whether that request had anything to do with this idea is not stated, so any link is an inference.

A slip in the book's own arithmetic. Soddy lists "recipes for the modern alchemist": thallium losing an alpha, bismuth losing two alphas, "or lead an a- and a beta-particle", would each give gold [read: Interp. printed p.233, PDF p.281]. Thallium (81 - 2 = 79) and bismuth (83 - 4 = 79) are right. Lead (82 - 2 + 1 = 81) gives thallium, not gold [calc, arbiter]. The man who formulated the displacement rules slipped while applying them, which is a reminder to check every step.

5. Worked example: the energy budget of one gram of radium

Question. Soddy says a gram of radium gives about 133 calories per hour and about 2.9 x 10^9 calories over its "complete change", "well over a million times" what burning coal gives. Can we rebuild those numbers from the atom up, see where his 2,500-year life came from, and see why the 1911 book used a smaller number? Everything below is reproduced by example.py (run command in its header). Modern nuclear data are [memory]: half-lives and alpha Q-values high confidence, beta Q-values medium.

Step 1. Atoms and activity

One gram of Ra-226 contains

N = N_A / 226.025 = 6.022 x 10^23 / 226.025 = 2.664 x 10^21 atoms.

The modern half-life is 1,600 years, so the mean life is T = 1,600 / 0.6931 = 2,308 years = 7.28 x 10^10 s. The activity is

lambda N = N / T = 2.664 x 10^21 / 7.28 x 10^10 s = 3.66 x 10^10 decays per second.

That is about one curie, as it should be, since the curie was defined from radium. With Soddy's 2,500-year mean life the activity would be 3.38 x 10^10 per second, 7.7% lower.

Step 2. Where Soddy's 2,500 years came from

Soddy measured the radium's life through its daughter. At equilibrium a gram of radium holds 0.6 mm^3 of emanation (radon) at STP [read: Interp. PDF p.123], and the emanation's mean life is 481,250 s = 5.57 days [read: p.157]. The equilibrium rule from section 3.2 gives T_Ra = T_Rn x (N_Ra / N_Rn):

• emanation atoms: N_Rn = (0.6 x 10^-3 cm^3 / 22,414 cm^3/mol) x N_A = 1.61 x 10^16
• ratio: N_Ra / N_Rn = 2.664 x 10^21 / 1.61 x 10^16 = 1.65 x 10^5
• T_Ra = 1.65 x 10^5 x 5.57 days = 2,520 years.

That matches the 2,500 years he adopted [calc; deep read]. Two cautions:

• The book gives no raw data, so the 0.6 mm^3 may itself have been worked out from the adopted 2,500 years. If so, this shows only that his numbers are consistent with each other, not that two independent measurements agree [deep read; arbiter section 6].
• Soddy says the first Ramsay-Soddy measurement of 1904 gave 1,150 years because too much gas was collected. He states the direction of the error himself [read: p.163].

With modern lives, the equilibrium volume should be about 0.65 mm^3, so his 0.6 is within 8% [calc].

Step 3. Heat from the decay energies

At equilibrium every member of the quick chain decays at the same rate as radium, 3.66 x 10^10 times a second. Each radium decay therefore brings with it one decay of each quick product:

Power = activity x energy per decay. Convert MeV to joules (x 1.602 x 10^-13) and watts to cal/h (x 3,600 / 4.184):

Soddy's 133 cal/h lies inside the 123-145 range. (The arbiter quotes 123-147; its upper end corresponds to 29.2 MeV per decay. The six Q-values above, which already include both beta decays, sum to 28.70 MeV and give 145. The difference changes nothing: 133 is inside either range.) A calorimeter absorbs all the alpha energy, most of the beta energy and only part of the gamma energy, and it never catches the antineutrinos, so a measured value should fall below 145 [calc; arbiter]. His split, "thirty-three calories per hour" from radium freed of emanation and "one hundred" from the emanation and its products [read: Interp. PDF pp.125-126], fits as well. Radium alone gives 24.6 cal/h, close to the 1909 edition's "25 from radium" [web: https://archive.org/details/interpretationof00soddrich]. By total Q, radium's own share is 17% against Soddy's 25%. The 1911 statement that the alpha rays carry "over 95% of the energy" [read: ME PDF p.210] is a little high: they carry 85% of the total Q [calc]. The experiment behind the split is Rutherford and Barnes (1903-04). They drove the emanation off radium bromide and watched the heat go with the gas, decay as the gas decayed, and come back as the emanation regrew in the radium [web, abstract level: https://www.nature.com/articles/068622a0]. That result weighs against the rival idea of the time, that radium soaks up energy from outside (Kelvin, Armstrong, L. Poincaré; [web, snippets via the Investigator]): an outside supply would have no reason to follow a gas that is carried into another vessel.

Step 4. The 1911 number: thirty-eight hours

Matter and Energy prints no calories-per-hour figure. It says the heat of one gram decomposes "in thirty-eight hours, its own weight of water into hydrogen and oxygen" [read: ME PDF p.210]. Splitting liquid water takes 285.8 kJ per mole [memory, high], which is 285,800 / 18.015 / 4.184 = 3,792 cal per gram. Spread over 38 hours, that is 99.8 cal/h. The 1911 book is therefore working with 100 cal/h, which the 1909 first edition of Interpretation prints outright [web: archive.org, as above]. The same number fits the other 1911 sentences: freezing to boiling "every hour" (100 cal), and "a quarter of a ton of coal" over the radium's life (step 6). So between 1909-11 and 1920 Soddy raised his heat constant from 100 to 133. That was a correction toward the modern range, not an arbitrary change: 100 was 25-30% low [arbiter].

Step 5. Total energy, and why "mean life" is the right multiplier

Soddy's arithmetic [read: Interp. PDF p.164]: 133 cal/h x 8,760 h per year = 1.16 x 10^6 cal per year, and 1/2,500 of the radium changes each year, so the total is 1.16 x 10^6 x 2,500 = 2.91 x 10^9 cal.

Why multiply by the mean life? The power falls as P(t) = P_0 e^(-t/T), so the total energy is the integral from 0 to infinity of P_0 e^(-t/T) dt = P_0 T (SymPy confirms this in example.py). The mean life is exactly right. Someone who multiplied by the half-life would get 2.0 x 10^9 cal, 1.44 times too small. Soddy never explains this, but his arithmetic is correct.

The per-atom count gives the same result with no lifetime at all. Total energy = N x energy per atom = 2.664 x 10^21 x 28.70 MeV = 1.225 x 10^10 J = 2.93 x 10^9 cal. The lifetime cancels because P_0 T = (N/T) x E x T = N x E. This also means the "complete change" total counts only the decays down to Po-214. The later Pb-210, Bi-210 and Po-210 decays, spread over decades, would add about 23% [calc, contrarian; Q-values medium confidence].

There is one subtlety. Soddy's two inputs each carry an error, and the errors happen to cancel. His mean life is 8% too long, and 133 cal/h is below the full Q (it implies 26.4 MeV absorbed per decay at the modern activity). Multiply 133 cal/h by the modern mean life of 2,308 years and you get 2.69 x 10^9 cal, about 8% less than he printed [calc].

Step 6. "A million times coal", and where the million comes from

Coal at 8,000 cal per gram [read: p.164]:

• radium / coal = 2.91 x 10^9 / 8,000 = 3.6 x 10^5
• coal burns with 32/12 = 2.67 g of oxygen per gram of carbon, so per gram of coal plus oxygen: 8,000 / 3.667 = 2,182 cal/g (Soddy prints 2,200)
• radium / (coal + oxygen) = 1.33 x 10^6, "well over a million" [calc].

Soddy prints all three bases: nearly a million against the most energetic chemical change, well over a million against coal plus oxygen, and 360,000 against coal alone (p.219). His "nearly a million times" compares radium with "a similar weight of matter undergoing any change known previously" [read: Interp. PDF p.164], not with coal alone. The most energetic familiar chemical change per gram of reactants is hydrogen burning to water, 3,792 cal/g (step 4), and 2.91 x 10^9 / 3,792 = 7.7 x 10^5, "nearly a million" [calc]. An earlier version of this primer (and all three reviewers) read the phrase as a comparison with coal alone and called it generous; that was a misreading, corrected by the QE [qe.md section 4 item 3; arbiter section 10.1]. Soddy's three figures are consistent with each other.

Where does the "million" come from? Compare the fraction of mass turned into energy:

• radium chain: 28.70 MeV / (226 x 931.5 MeV) = 1.36 x 10^-4
• burning carbon: 393.5 kJ per mole of CO2 / (0.044 kg x c^2) = 9.9 x 10^-11 [memory, high for the heat of combustion]
• ratio: 1.4 x 10^6 [calc].

So "a million times coal" is the ratio of nuclear to chemical binding energies per unit mass. Over its whole life (to Po-214), one gram of radium loses 0.136 milligrams of mass. Soddy had the size right without the formula that explains it.

What the example teaches

• The numbers in Soddy's books hang together, and they agree with modern nuclear data to within about 10%.
• The two places where they are "wrong" (100 cal/h in 1911; mean lives 8-24% long) are measurement limits of 1909-1920, and Soddy corrected the first of them himself.
• The thing he could not supply, why the energy is so large, is E = mc^2 applied to nuclear masses. The million-fold factor is simply the ratio of nuclear to chemical mass defects.

6. Evidence table

Verdicts are the arbiter's. "Tested?" means tested experimentally by anyone, then or since.

No claim here is "Never properly tested". Nothing in these books was dismissed without a test. The nearest case is C8b (living cells as molecular sorters): no experiment was ever designed for that exact proposal, and the arbiter scores it on the general record of physiology and thermodynamics. The disintegration theory itself was fought openly. Kelvin campaigned against it in The Times in August 1906, and Soddy answered in Nature (vol. 74, pp.516-517): disintegration "is based on experimental evidence, which even its most hostile opponents are unable to shake" [web: research/files/nature-1906-...pdf, via the Investigator]. Experiment settled it within about a decade. Kelvin's own letters were not read for this review, so his side is known here only through Soddy's reply and press snippets.

7. What a learner should take away

What is solid. - The exponential decay law, mean life = 1/lambda, and the factor 1.44 between mean life and half-life. - Secular equilibrium (amount proportional to mean life inside a chain) and its use to measure lives too long to watch. - The displacement rules (alpha: Z - 2, A - 4; beta: Z + 1) and the isotope concept. - That radioactive heat is internal and about a million times chemical energy per gram. Every one of these came from experiment and has survived a century of testing. - Soddy's arithmetic in the key passages is right, including the subtle choice of the mean life (not the half-life) as the multiplier for total energy.

What went wrong, and why each mistake is instructive. - Gold and silver. He used a rule (abundance proportional to life) outside the setting where it holds (members of one chain with a long-lived parent). Always check a rule's conditions before applying it. - The heating, exploding Earth. He took Joly's uniform-radium, conduction-only model on trust. A single order-of-magnitude check against the measured heat flow, which was not available in this form in 1920, rules it out by a factor of about 40. The physics of the source was right; the model of how heat gets out was wrong. - The proton radius. Applying the classical "electromagnetic mass" formula to the proton silently assumes that all its mass is electrostatic. A hidden assumption gave an answer about 550 times too small. - Two to four million volts. A classical barrier argument overestimates what is needed, because quantum particles tunnel. The direction (atomic energy is there) was right; the route was wrong. - Small slips: "58 grams" where his own table gives 40; "lead an alpha and a beta" gives thallium, not gold; "uranium I, with its period of 8,000,000 years", three zeros short [read: page image Interp. PDF p.237]; uranium II as "about two million years" in the text and "3,000,000 years (?)" in the tables [read: Interp. PDF pp.195, 199, 208, 211]. Matter and Energy misprints the 1890s dates as 1905 in four places: the inert gases "discovered in 1905" and "before 1905 only one monatomic gas was known" (argon 1894, terrestrial helium 1895), and the X-rays "until 1905" and "by Röntgen in 1905" (1895) [read: ME PDF pp.128 (twice), 165, 204; qe.md]. Careful authors make these too, so check the arithmetic yourself. - Text carried forward between editions. The helium-from-uranium agreement and the "thirteen per cent" were written for the 1909 edition, when the uranium chain had fewer alpha emitters, and were not updated in 1920 when the chain grew (section 4). The 1920 preface says the text was "amplified rather than modified" [arbiter section 10.3]. A new edition is a good place to look for statements that no longer fit the new material.

What is genuinely open (mostly history, not physics). - The strict constancy of decay rates at the level of parts per thousand. A small literature claims solar or seasonal modulation and the better-controlled measurements do not find it. Long-term, multi-site, shielded counting would decide it. - Priority for the 1913 displacement law. The 1920 book credits "A. S. Russell, K. Fajans, and the writer" in "the same month" [read: Interp. PDF p.277]. The 1922 Nobel lecture adds Hevesy and gives an order: Hevesy and Russell partly correct, Fajans first complete, Soddy a fortnight later and independently [web: https://www.nobelprize.org/uploads/2018/06/soddy-lecture.pdf, printed p.387]. Both are Soddy's own accounts, and he had a stake in them. The 1920 book also credits two others: Alexander Fleck, who "spent three years in the exhaustive study of the chemical nature of all the radioactive elements" in Soddy's Glasgow laboratory and supplied much of the chemistry behind the rules [read: Interp. PDF p.276], and van den Broek, whose suggestion that successive places in the periodic table differ by one unit of nuclear charge "was made tentatively by van der Broeck [sic] before it was first proved by these researches" [read: p.279]. The word "isotope" was suggested to him by Margaret Todd in 1913 [web: https://en.wikipedia.org/wiki/Margaret_Todd_(doctor)]. The deciding records are the dated 1913 papers, the Fajans letter of 19 October 1913 and the 1935 correspondence with W. A. Noyes in the Bodleian catalogue [web: research/2026-10-08-archiveshub-csac-catalogue.md]. None of them has been read. - Radium supply and the mesothorium patent. Soddy held a 1910 British patent (GB 25,504) for extracting mesothorium from monazite [web: https://patents.google.com/patent/GB191025504A/en]. A Springer chapter snippet says the patent threatened the German firm Knoefler & Co., and New York Times snippets of November-December 1913 report Soddy calling for radium supply to be nationalised [web, snippets via the Investigator; articles not opened]. Soddy was both patent holder and campaigner here, an interested party, and the German firm's side is absent from the record. The Glasgow and West of Scotland Radium Committee minute book (University of Glasgow Archives) would show what happened [arbiter section 8C]. Not a physics question. - Gaps in Soddy's papers (absence of evidence). The Bodleian catalogue says his Aberdeen war-work envelopes were "probably ... disturbed, and some of them removed". The originals of his 1953-54 letters to Muriel Howorth are lost; only Trenn's 1969 copies of 32 survive. His secret 1916 patent 5527/16 (fireless boilers) was not found in Google Patents [web: research notes; arbiter section 8]. The record allows several explanations: - ordinary loss or re-sorting by his literary trustees (the catalogue's "probably", and the 18 years between his death and the cataloguing); - removal of classified wartime material (the topic was Admiralty war work); - selection by Howorth for her 1958 biography; - deliberate suppression for his politics (he attacked banks and private radium supply, and late in life the hydrogen-bomb tests).

No document supports the last explanation, and the first three explain the same gap. The records that would decide are the Admiralty Board of Invention and Research files at the National Archives (not yet searched), the Patent Office record of 5527/16, and Howorth's papers. None of this touches the physics. - His foresight about atomic energy. H. G. Wells dedicated The World Set Free (1914) to The Interpretation of Radium and wrote that it "owes long passages to the eleventh chapter" [web: https://www.gutenberg.org/cache/epub/1059/pg1059.txt]. Guston (2012) documents that Soddy warned publicly of atomic war in 1915 [web: https://pmc.ncbi.nlm.nih.gov/articles/PMC3456912/]; the 1915 text itself was not found. The Times reported Rutherford in 1933 as calling the hope of power from atoms "moonshine" [web: https://nuclearfaq.ca/yonge.htm, a popular article; the wording is the newspaper's]. On this question the senior authority was more cautious than Soddy, and Soddy's general direction proved right. His unread 1946 manuscript "Real Alchemy and the Release of Atomic Energy" and his 1952 and 1954 Lindau lectures would show how his views developed. - How "energy and wealth" became an economics. Soddy's later monetary books (outside the folder) are cited as a starting point of ecological economics (Daly 1980; Merricks 1996) [web, snippets via the Investigator]. A reader who wants to weigh that half of his work has to read them; this review did not.

8. Check yourself

1. Radon-222 has a half-life of 3.82 days and radium-226 a half-life of 1,600 years. In an old radium sample at equilibrium, how many radon atoms are there per radium atom? Does it matter whether you use half-lives or mean lives here?
2. Soddy multiplies the heat per year by 2,500 years to get the total heat from a gram of radium. Show with an integral why the mean life, not the half-life, is the right multiplier. By what factor would the half-life version be wrong?
3. Soddy wrote that lead losing "an alpha and a beta particle" would give gold. What does it actually give? Using only alpha and beta-minus decays, what is the shortest recipe from lead-206 (Z = 82) to an isotope of gold (Z = 79), and what is the mass number of the gold?
4. Joly's model puts 2 x 10^-12 g of radium in every gram of the Earth (mass 5.97 x 10^27 g). Radium with its quick products gives about 0.16 W per gram. Estimate the total heat output in terawatts and compare it with the measured 44-47 TW. What does that tell you about the model?
5. Soddy found that the actinium series in uranium ores gives 1/15 to 1/16 of the uranium series' alpha activity. Read it the modern way: the actinium series comes from uranium-235 (half-life 704 million years, 7 alpha decays down to lead-207), the main series from uranium-238 (half-life 4,468 million years, 8 alphas). What fraction of uranium atoms would be U-235 if his ratio were exact? Compare with the true 0.72%.

9. Further reading

Mainstream textbooks. - Kenneth S. Krane, Introductory Nuclear Physics (Wiley, 1988): the chapters on radioactive decay (including equilibrium), alpha decay (Gamow's tunnelling theory, Geiger-Nuttall), beta decay, and nuclear masses. - Serway, Moses and Moyer, Modern Physics: the nuclear-physics chapters, at the reader's level. - David J. Griffiths, Introduction to Quantum Mechanics: the WKB chapter, worked example on alpha decay. - Turcotte and Schubert, Geodynamics: the Earth's heat flow and why mantle convection matters (for claim 7a).

History. - Abraham Pais, Inward Bound: Of Matter and Forces in the Physical World (1986): the history of radioactivity and the nucleus [memory, high]. - Thaddeus J. Trenn, The Self-Splitting Atom: The History of the Rutherford-Soddy Collaboration (1977) [memory, high; named in the Investigator's leads]. - Linda Merricks, The World Made New: Frederick Soddy, Science, Politics, and Environment (1996) [web snippets via the Investigator]. - Lawrence Badash, writings on the radium-energy controversy (1966 and later) [web snippets via the Investigator]. - Mark Morrisson, Modern Alchemy: Occultism and the Emergence of Atomic Theory (Oxford University Press): the setting of Soddy's "forgotten race" passage [web snippets; not opened]. - Richard Rhodes, The Making of the Atomic Bomb (1986): Wells, Soddy and Szilard [memory, high]. - David H. Guston, "The Pumpkin or the Tiger? Michael Polanyi, Frederick Soddy, and Anticipating Emerging Technologies", Minerva (2012) [web: https://pmc.ncbi.nlm.nih.gov/articles/PMC3456912/].

Soddy's own works. - The Interpretation of Radium and the Structure of the Atom, 4th ed. 1920 (in the folder). The 1909 first edition is on archive.org (interpretationof00soddrich) [web]. - Matter and Energy (1911) (in the folder). - Nobel lecture, "The origins of the conceptions of isotopes", 12 December 1922 [web: https://www.nobelprize.org/uploads/2018/06/soddy-lecture.pdf]. - "The Recent Controversy on Radium", Nature 74, 516-517 (1906): his reply to Kelvin [web: research/files]. - Cartesian Economics (1922) and Wealth, Virtual Wealth and Debt (1926), on archive.org [web]: his energy-based economics, not reviewed here. - H. G. Wells, The World Set Free (1914), Project Gutenberg #1059: the novel built on chapter XI of the Interpretation.

---

Answers

1. At equilibrium N_Rn / N_Ra = T_Rn / T_Ra. The ratio of mean lives equals the ratio of half-lives, because each mean life is its half-life divided by the same ln 2. So either works: 3.82 / (1,600 x 365.25) = 6.5 x 10^-6 radon atoms per radium atom. One gram of radium (2.66 x 10^21 atoms) holds 1.7 x 10^16 radon atoms, about 0.65 mm^3 of gas at STP, close to Soddy's 0.6.

2. The power is P(t) = P_0 e^(-t/T). The total is the integral from 0 to infinity of P_0 e^(-t/T) dt = P_0 T. With the half-life t_1/2 = 0.693 T you would get 0.693 P_0 T, too small by 1/ln 2 = 1.44. Physically: each atom lives on average T, not t_1/2, so the average atom contributes for a time T.

3. Lead (82) minus an alpha (80, mercury) plus a beta (81) is thallium, not gold. To reach 79 with alphas (-2 each) and betas (+1 each) you need 2a - b = 3, and the smallest solution is two alphas and one beta: 82 - 4 + 1 = 79. The mass number is 206 - 8 = 198, which is gold-198 (radioactive, half-life about 2.7 days [memory, high]). The only stable gold is 197, which this recipe cannot reach from lead-206.

4. Radium mass = 5.97 x 10^27 x 2 x 10^-12 = 1.19 x 10^16 g. Heat = 1.19 x 10^16 x 0.164 W = 1.96 x 10^15 W = about 1,960 TW, some 40 times the measured 44-47 TW. A uniform distribution of radium at that concentration would pour out far more heat than the Earth actually loses. Radioactive elements must be concentrated near the surface, where their heat escapes (as Strutt's crustal measurements suggested), and the deep interior cannot be heating up the way Joly's model required [calc, arbiter].

5. At equilibrium every member of a series has the activity A of its parent, so the alpha activity ratio is r = 7 A_235 / (8 A_238), and A_235 / A_238 = 8r/7 = 0.071 to 0.076. Activity is A = N / T (T the mean life), so N_235 / N_238 = (A_235 / A_238) x (T_235 / T_238), and the ratio of mean lives equals the ratio of half-lives, 704 / 4,468 = 0.158. That gives N_235 / N_238 = 0.011 to 0.012, so about 1.1-1.2% of uranium atoms, against the true 0.72%. Run backwards, the true 0.72% gives r = 0.040, about 1/25. So Soddy's measured ratio was about 1.6 times too high, but it was the right order, and the separate-isotope reading explains it without any branch [calc: example.py, last section; arbiter section 10.2].

---

Corrections to earlier documents (for the orchestrator)

1. arbiter.md, section 1 "Numbers I recomputed", and section 4 C2 ("123-147 cal/h"). The upper end should be about 145 cal/h, not 147. The six Q-values of the quick chain (Ra-226 4.871, Rn-222 5.590, Po-218 6.115, Pb-214 1.018, Bi-214 3.270, Po-214 7.833 MeV; [memory], high for alphas, medium for betas) sum to 28.70 MeV, and 3.66 x 10^10 decays/s x 28.70 MeV = 0.168 W = 144.7 cal/h [calc: example.py, step 3]. The arbiter's 147 corresponds to 29.2 MeV, which was carried over from the mathematician's "about 29 MeV ... 29.2" and is about 0.5 MeV too high. The per-atom total is then 2.93 x 10^9 cal (not 2.98 x 10^9). Small, and it changes no verdict: 133 cal/h still lies inside the range. Any copy of "123-147" (arbiter, entries) can read "123-145".
2. Reader note: Interp. PDF p.125 gives the 1920 split as "thirty-three calories per hour" from radium freed of emanation and "one hundred" from the emanation and its products. This is quoted in the Matter and Energy deep read's comparison table; it backs the 25:75 split in the 1920 book as well as the 1909 one. No change needed, recorded for completeness. (The split runs from p.125 to p.126; the citation in step 3 now reads pp.125-126, per QE correction 8.)
3. arbiter.md section 10.6 item 18 (the 123-147 range). The arbiter keeps 123-147 and says the upper end is "145 with the primer's Q-value set (28.7 MeV) and 147 with 29.2 MeV including beta Q-values". That rationale does not hold: the primer's 28.70 MeV already includes both beta Q-values (Pb-214 1.018 and Bi-214 3.270 MeV), so 29.2 MeV is not "28.7 plus the betas"; it is about 0.5 MeV above the sum of the six Q-values, and no Q-value set behind it has been shown [calc: example.py, step 3]. The primer uses 123-145 and notes the arbiter's 147 beside it. No verdict changes.
4. arbiter.md section 10.3, C1, pleochroic halos "[read: Interp. PDF pp.165-167]". Those are printed page numbers; the PDF pages are 211-215 (deep read heading: "printed pp.161-167; PDF pp.207-215"; qe.md section 3 item 5). The primer cites both.
5. arbiter.md section 10.6 item 22 (does the primer present the helium measurement as independent confirmation?). Checked: the earlier primer did not mention the helium measurement at all. It is now in section 4 and the evidence table as Internally inconsistent within the 1920 edition, showing the order of magnitude only.
6. This primer's own earlier step 6 sentence ("His phrase 'nearly a million times' for coal alone is generous") was a misreading of Interp. PDF p.164, carried from the reviews; replaced with QE correction 3 (step 6 now states all three bases). Its earlier evidence-table wording ("'million' is per gram of coal plus oxygen") and section 1 wording ("a million times ... any chemical fuel") are corrected the same way. Copies of the old wording may sit in work/rankings/entries/Soddy.json and site/; the orchestrator's stale-copy search should include "generous".