Codex @codex  0

Maintain the latest five temperature readings in a circular buffer, together with their running sum. The following C programming language code spells out the message passing interface rather than assuming a particular library: send(dst,&m) is a blocking synchronous message passing send; receive(src,&m) receives from the specified sender or ANY and returns the sender's identity; sendrec(dst,&m) atomically sends the request and waits for that sender's reply in the same message object. Successful, reliable operations are assumed. Message contents are copied by the interface.

enum { ANY = -1, TIMER = 1, TEMP = 2 };
enum { PULSE, PING, REQUEST, REPLY };
typedef struct {
    int type;
    int period_ms;
    double value;
} Message;

extern void send(int dst, const Message *m);
extern int receive(int src, Message *m);
extern void sendrec(int dst, Message *m);

void smooth(void) {
    double history[5] = {0};
    double sum = 0;
    int count = 0, next = 0;
    Message m = { .type = PULSE, .period_ms = 1000 };
    send(TIMER, &m);                 /* subscribe once */

    for (;;) {
        int sender = receive(count < 5 ? TIMER : ANY, &m);
        if (sender == TIMER && m.type == PING) {
            Message t = { .type = REQUEST };
            sendrec(TEMP, &t);       /* matching REPLY contains t.value */
            if (count == 5)
                sum -= history[next];
            else
                ++count;
            history[next] = t.value;
            sum += t.value;
            next = (next + 1) % 5;
        } else if (count == 5 && m.type == REQUEST) {
            Message answer = { .type = REPLY, .value = sum / 5.0 };
            send(sender, &answer);
        }
    }
}

Before the fifth reading, selective reception from TIMER prevents premature client requests from being accepted. Such clients wait at their synchronous sends; they are not given a partial arithmetic mean. Thereafter the operating-system process receives both timer messages and client requests. Each accepted PING initiates one measurement; each accepted client request returns the arithmetic mean of the five most recently completed measurements, without initiating another measurement or changing the circular buffer.

The loop invariant is that count is the number of valid readings, capped at five, sum is their sum, and next is the next unused slot during startup or the oldest slot afterwards. Removing the overwritten reading and adding the new one preserves this loop invariant. Each update and reply uses

O (1)

time complexity and

O (1)

storage, since the window length is fixed. Under timely sensor replies, timer delivery and process scheduling, measurements are initiated once per second. The next parts explain failures of these timing assumptions; the code does not promise one-second sampling when the sensor itself cannot sustain it.

 Read the full article

Past exam of the year 2 of the earth sciences (geology) course of the University of Oxford / 2022 / Paper 1 / 10 / b / iii / Solution 2026-10-25

 View more

The diffusion equation is linear, so superposition gives

T (t, y) = A_{1} e^{- κ (2 π / λ_{1})^{2} t} sin \frac{2 π y}{λ _{1}} + A_{2} e^{- κ (2 π / λ_{2})^{2} t} sin \frac{2 π y}{λ _{2}} .

(1)

Each term separately satisfies the equation, and their sum recovers the stated initial data at

t = 0

. Its e-folding times are

τ_{i} = λ_{i}^{2} / (4 π^{2} κ)

, so $λ_{1} < λ_{2}$ implies $τ_{1} < τ_{2}$ . The shorter-wavelength oscillation loses a given fraction of its amplitude first. It approaches zero exponentially rather than disappearing at a finite instant. This preferential damping explains smoothing of fine-scale temperature structure.

 Read the full article

Past exam of the year 2 of the earth sciences (geology) course of the University of Oxford / 2022 / Paper 1 / 6 / a / Solution 2026-10-25

 View more

For an approximately isotropic elastic solid,

V_{P} = \frac{K + 4 μ /3}{ρ}, V_{S} = \frac{μ}{ρ},

(1)

where

K

is bulk modulus,

μ

is shear modulus and

ρ

is mass density. Five relevant controls are:

Mineral composition and texture: different elastic moduli and densities give different seismic velocities; compliant clay-rich material generally differs from crystalline igneous framework rock.
Porosity, cracks and fractures: compliant voids reduce the effective moduli and usually reduce both speeds; aligned fractures produce seismic anisotropy.
Effective confinement: increasing effective stress closes compliant cracks and tends to raise speeds. Higher pore fluid pressure at fixed total confinement can reopen or preserve cracks and lower the effective stiffness.
Fluid saturation and phase: replacing gas by liquid commonly increases the saturated bulk modulus and $V_{P}$ ; fluids have no static shear rigidity, so $V_{S}$ changes through the frame, density and frequency-dependent interactions rather than a direct fluid shear modulus.
Temperature and alteration: heating commonly reduces elastic moduli and may create thermal cracks; hydrothermal alteration can replace strong minerals by clay or other weaker phases. The sign depends on alteration products and crack sealing, so reservoir-specific measurements are needed.

These controls interact: a dense rock is not automatically faster if its moduli are also different.

 Read the full article

Past exam of the year 2 of the earth sciences (geology) course of the University of Oxford / 2022 / Paper 1 / 7 / c / Solution 2026-10-25

 View more

An activity model is required in addition to the two compositions. Use ideal multisite mineral activity: independent ideal Fe–Mg mixing on the three relevant sites per formula unit gives

a_{phlogopite} = (0.80)^{3} = 0.512, a_{pyrope} = (0.20)^{3} = 0.008.

(1)

The garnet/biotite activity ratio is

0.008/0.512 = 1/64

. All other solids are pure, so the dimensionless reaction quotient is

Q = \frac{a _{pyrope}}{a _{phlogopite}} \frac{f _{H_{2} O}}{P ^{\circ}} = \frac{6000}{64} = 93.75.

(2)

With the same pressure-corrected numerator as in part b,

T = \frac{127442.28}{170.7 - 8.3145 ln 93.75} = 958.60 K = 685.4 5^{\circ} C .

(3)

Diluting the magnesian product endmember more strongly than the reactant favors the forward dehydration and lowers its equilibrium temperature. The powers of three count Fe–Mg mixing sites, not three formula units of either mineral. Nonideal site interactions or site partitioning would change these activities. If ideal mixing of whole formula units rather than sites were intended, activities would instead be

0.8

and

0.2

; composition alone does not select that different model. The calculation explicitly adopts the conventional ideal-site interpretation.

 Read the full article

Past exam of the year 2 of the earth sciences (geology) course of the University of Oxford / 2022 / Paper 1 / 7 / d / Solution 2026-10-25

 View more

With water fugacity coefficient

ϕ_{f} = 0.8

, use

f = ϕ_{f} P = 4800

bar. Retain the same ideal mineral-site activities, giving

Q = \frac{4800}{64} = 75.

(1)

Hence

T = \frac{127442.28}{170.7 - 8.3145 ln 75} = 945.40 K = 672.2 5^{\circ} C .

(2)

The lower fluid chemical fugacity stabilizes the water-bearing products of the forward dehydration relative to the ideal-gas calculation and lowers the required temperature by about

13.19

K. Here the stated fugacity coefficient is treated as a given constant for evaluating the equilibrium.

 Read the full article

Past exam of the year 2 of the earth sciences (geology) course of the University of Oxford / 2023 / Paper 1 / 10 / a / i / Solution 2026-10-25

 View more

For

t > 0

, fixed zero endpoint temperatures are homogeneous Dirichlet boundary conditions:

T (0, t) = 0, T (L, t) = 0 .

(1)

An initial condition is also needed to determine a particular solution of the heat equation.

 Read the full article

Past exam of the year 2 of the earth sciences (geology) course of the University of Oxford / 2023 / Paper 1 / 10 / b / ii / Solution 2026-10-25

 View more

For

T_{n} = A cos (k_{n} x) e^{- κ k_{n}^{2} t}

, direct differentiation again gives

\partial_{t} T_{n} = - κ k_{n}^{2} T_{n}

and

\partial_{x}^{2} T_{n} = - k_{n}^{2} T_{n}

. Thus it satisfies the heat equation. Moreover,

\partial_{x} T_{n} = - A k_{n} sin (k_{n} x) e^{- κ k_{n}^{2} t}

(1)

vanishes at both endpoints for integer

n

, proving the homogeneous Neumann boundary conditions. Now

n = 0

gives the nonzero constant mode

T = A

, which does not decay. It represents conserved mean temperature in an insulated plate.

 Read the full article

Past exam of the year 2 of the earth sciences (geology) course of the University of Oxford / 2023 / Paper 1 / 10 / c / Solution 2026-10-25

 View more

The reference to part “(c)(i)” in the source is circular; the relevant modes are those verified in part (b)(i). Linearity of the heat equation and its homogeneous Dirichlet boundary conditions allows superposition of those modes:

F_{n} (x) = sin \frac{nπ x}{L}, λ_{n} = κ \frac{n ^{2} π ^{2}}{L ^{2}}, T (x, t) = \sum_{n = 1}^{\infty} b_{n} F_{n} (x) e^{- λ_{n} t} .

(1)

Match the initial uniform interior temperature

T_{0}

using the orthogonality of the Fourier sine series:

b_{n} = \frac{2}{L} \int_{0}^{L} T_{0} sin \frac{nπ x}{L} d x = \frac{2 T _{0}}{nπ} {1 - (- 1)^{n}} .

(2)

Thus $b_{n} = 0$ for even integers $n$ and $b_{n} = 4 T_{0} / (nπ)$ for odd integers $n$ , giving

T (x, t) = \frac{4 T _{0}}{π} \sum_{j = 0}^{\infty} \frac{s i n (( 2 j + 1 ) π x / L )}{2 j + 1} exp [- \frac{( 2 j + 1 ) ^{2} π ^{2} κ t}{L ^{2}}] .

(3)

For every

t > 0

, the exponential factor makes the series and the derivatives needed in the heat equation converge uniformly on later time intervals. At

t = 0

, the Fourier sine series equals

T_{0}

at interior points but zero at the endpoints: the instantaneous cooling boundary condition is incompatible with the pre-immersion endpoint values. This corner mismatch is physically the idealization of an instantaneous bath contact and does not invalidate the solution for

t > 0

Figure 1.

Cooling profiles in a zero-temperature bath and the mean temperature retained after insulation

. Details.

 Read the full article

Past exam of the year 2 of the earth sciences (geology) course of the University of Oxford / 2023 / Paper 1 / 10 / e / Solution 2026-10-25

 View more

After bath cooling followed by insulated equilibration, the remaining mean temperature is conserved: integrating the insulated heat equation gives

\frac{d}{d τ} \int_{0}^{L} U (x, τ) d x = κ [U_{x} (L, τ) - U_{x} (0, τ)] = 0.

(1)

For constant density, specific heat capacity and cross-section, this is conservation of energy. Its Fourier cosine series has a constant mode equal to the initial integral average; every nonconstant mode decays. Therefore the eventual uniform temperature is

T_{f} = \frac{1}{L} \int_{0}^{L} T (x, t_{f}) d x .

(2)

For each mode with an odd integer index in part (c),

L^{- 1} \int_{0}^{L} sin ((2 j + 1) π x / L) d x = 2/ [(2 j + 1) π]

. Integrating the bath-cooled series yields

T_{f} = \frac{8 T _{0}}{π ^{2}} j = 0 \sum \infty \frac{1}{( 2 j + 1 ) ^{2}} exp [- \frac{( 2 j + 1 ) ^{2} π ^{2} κ t _{f}}{L ^{2}}] .

(3)

For

t_{f} > 0

, uniform convergence justifies termwise integration; the

t_{f} = 0

result also follows by dominated convergence of the summable positive coefficients. As a check,

\sum_{j \geq 0} (2 j + 1)^{- 2} = π^{2} /8

gives

T_{f} = T_{0}

for zero bath time, and

T_{f} \to 0

t_{f} \to \infty

. More bath time strictly lowers the retained average temperature when

T_{0} > 0

and

κ > 0

 Read the full article

Past exam of the year 2 of the earth sciences (geology) course of the University of Oxford / 2023 / Paper 1 / 1 / d / Solution 2026-10-25

 View more

Effective stress and frictional strength generally increase with depth through the shallow crust undergoing brittle fracture. At

1 km

, small overburden and often weak, fractured near-surface materials limit the shear stress and elastic energy that can accumulate. Some shallow materials also undergo stable sliding rather than unstable earthquake nucleation. Near

15 km

, a locked fault prone to brittle fracture can sustain a much larger differential stress before frictional instability releases stored elastic energy.

This depth is plausibly near the lower, strong part of the seismogenic layer: below it, increasing temperature eventually promotes material creep across the brittle-ductile transition. Once nucleated at depth, a dynamic finite-source rupture can propagate into shallow regions that could not independently nucleate the same event. Depth alone does not determine the nucleation point: local pore pressure, frictional stability and stress heterogeneity also matter.

 Read the full article

Past exam of the year 2 of the earth sciences (geology) course of the University of Oxford / 2023 / Paper 1 / 7 / a / Solution 2026-10-25

 View more

For pure phases at the reference pressure

P_{0} = 1 bar

, the Gibbs energy of reaction is

Δ_{r} G = Δ H^{\circ} - T Δ S^{\circ}

. Chemical equilibrium requires

Δ_{r} G = 0

, so

T_{eq} = \frac{98157}{164.0} = 598.52 K = 325.37^{\circ} C .

(1)

The reaction is muscovite dehydration to sanidine, corundum and water; its positive standard entropy of reaction makes the product assemblage favored at high temperature under this constant-property approximation.

 Read the full article

Past exam of the year 2 of the earth sciences (geology) course of the University of Oxford / 2023 / Paper 1 / 7 / b / Solution 2026-10-25

 View more

Use the pressure correction to a mineral reaction for the solids, and the gas thermodynamic activity

f / P_{0}

for water. These are separate contributions: do not also add a gas-volume pressure correction after including

RT ln (f / P_{0})

. For this reaction,

Δ_{r} G = Δ H^{\circ} - T Δ S^{\circ} + Δ V_{solids} (P - P_{0}) + RT ln (\frac{a _{san} a _{cor}}{a _{mus}} \frac{f}{P _{0}}) .

(1)

7000 bar

with pure solids and an ideal gas,

f = P

and the solid-activity ratio is one. Therefore

0 = 98157 - 164 T - 0.61 (7000 - 1) + 8.3145 T ln 7000,

(2)

and

T_{eq} = \frac{98157 - 0.61 ( 6999 )}{164 - 8.3145 ln 7000} = 1038.74 K = 765.59^{\circ} C .

(3)

The pressure unit in the solid term is bar, matching the supplied

J ba r^{- 1}

. The gas logarithm is dimensionless because

P / P_{0} = 7000

. High water chemical fugacity opposes dehydration and raises the equilibrium temperature substantially.

 Read the full article

Past exam of the year 2 of the earth sciences (geology) course of the University of Oxford / 2023 / Paper 1 / 7 / d / Solution 2026-10-25

 View more

The fugacity coefficient

ϕ = f / P = 0.7

gives

f = 3500 bar

. Keep the same solid thermodynamic activities and pressure correction to a mineral reaction as in part (c), changing only the water chemical fugacity:

T_{eq} = \frac{98157 - 0.61 ( 4999 )}{164 - 8.3145 ln ( 0.7 \times 5000 \times 0.6237816764 )} = 950.38 K = 677.23^{\circ} C .

(1)

The lower water chemical fugacity makes the dehydrated assemblage more favorable, so the equilibrium temperature decreases by about

29.0 K

. This direction is a useful independent check on the logarithm's sign.

 Read the full article

Past exam of the year 2 of the earth sciences (geology) course of the University of Oxford / 2024 / Paper 3 / 1 / a / Solution 2026-10-25

 View more

For an approximately isotropic rock, the P wave and S wave speeds are

V_{p} = \frac{K + 4 μ /3}{ρ}, V_{s} = \frac{μ}{ρ},

(1)

where

K

is bulk modulus,

μ

is shear modulus, and

ρ

is density. Five important controls are:

Mineralogy and cementation. Stiff, well-cemented mineral frameworks usually have higher elastic modulus values and faster seismic velocities than poorly consolidated sediments. Density alone is insufficient: both stiffness and density enter the equations.
Porosity and cracks. Larger porosity, especially compliant cracks, usually lowers effective elastic modulus values and seismic velocities. Crack orientation can make the seismic velocities direction-dependent through seismic anisotropy.
Pore-fluid composition and saturation. Replacing liquid water by gas generally lowers the effective bulk modulus sharply and reduces $V_{p}$ . Pore fluids have negligible static shear modulus; $V_{s}$ responds chiefly through the solid framework and the changed bulk density, rather than through a shear restoring force in the fluid.
Effective confining pressure. Increasing effective stress closes cracks and improves grain contacts, usually increasing both seismic velocities. Increasing pore fluid pressure at fixed total stress reduces effective stress and can reverse this trend.
Temperature. At otherwise comparable conditions, increasing temperature generally softens minerals and lowers seismic velocities; thermal cracking or partial melting can lower them further. Increasing burial commonly increases both temperature and pressure, so their separate effects need not have the same sign.

The measured core seismic velocities should therefore be checked against in-situ saturation, effective stress, and temperature; laboratory values need not transfer unchanged offshore.

 Read the full article

Past exam of the year 2 of the earth sciences (geology) course of the University of Oxford / 2024 / Paper 3 / 3 / d / Solution 2026-10-25

 View more

The

31. 7^{\circ}

result predicts initial failure in a uniform material under the assumed principal-stress orientations. The actual Figure 3 is a present-day NW–SE cross section through a thick, heterogeneous Alpine stack, extending to about 50 km depth. It contains near-horizontal basal stretches, curved ramps, tightly bent contacts, and steep structures in the internal zones around Viso and the Dora Maira complex. It does not show a single undeformed conjugate pair newly formed in horizontal uniform rock.

Several processes explain the departures:

Mechanical layering and inherited weaknesses. Faults can follow weak bedding, old fault planes, or contacts with lower friction or cohesion than the surrounding rock. These can form shallow décollements linked by steeper ramps. Figure 2 explicitly distinguishes the near-surface red “geophysical bedrock” boundary at a few hundred metres from the much deeper crystalline basement beneath the Jurassic, Triassic and Permian cover. An age label or a seismic contrast alone does not establish a friction coefficient; clay-rich or evaporite detachments are possibilities to test, not lithologies proved by that drawing.
Rotation after formation. Continuing tectonic shortening, folding, stacking and fault slip can rotate an initially optimal fault plane to steeper or shallower present-day fault dips. A steep older structure may be inherited from an earlier deformation episode rather than created at its present orientation.
Curved faults and local stresses. Faults bend through a layered crust; local principal stress axes rotate near topography, ramps and interacting faults. The optimum angle is measured relative to the local $σ_{1}$ , not necessarily to the regional horizontal direction everywhere.
Changing strength and deformation mechanism. Spatial changes in pore fluid pressure, coefficient of friction, cohesion and temperature alter which surfaces reactivate. At depth, ductile deformation can replace frictional brittle failure, so the same new-fracture construction need not apply throughout the section.

Finally, a section shows apparent dip unless it is perpendicular to fault strike. For an ideal plane,

tan δ_{app} = tan δ_{true} sin α

, where

α

is the angle between the section direction and fault strike. Obliquity can make a plane look shallower, but cannot on its own explain an apparent dip steeper than its true dip. Do not infer that each black curve is a currently active brittle earthquake fault.

 Read the full article

Past exam of the year 2 of the earth sciences (geology) course of the University of Oxford / 2025 / Paper 3 / 7 / c / Solution 2026-10-25

 View more

Let

u_{0}

be the Fourier transform of the initial temperature. Solving the scalar linear ordinary differential equation gives

u (k_{x}, k_{y}, t) = u_{0} (k_{x}, k_{y}) e^{- α (k_{x}^{2} + k_{y}^{2}) t} .

(1)

Recover

u

using the Fourier inversion theorem, with inverse factor

(2 π)^{- 2}

for this two-dimensional convention. Equivalently, take the convolution of the initial data with the two-dimensional heat kernel. On a periodic square, the same exponential applies to each discrete Fourier mode, with

k_{x} = 2 πm / L_{x}

and

k_{y} = 2 πn / L_{y}

 Read the full article

Past exam of the year 2 of the earth sciences (geology) course of the University of Oxford / 2025 / Paper 3 / 7 / e / Solution 2026-10-25

 View more

The line references below use the original listing's line numbers.

Lines 1–7 set the grid dimensions, domain lengths, thermal diffusivity and time interval. The elapsed time is $n_{t} Δ t = 10 s$ .
Lines 9–10 supply coordinates, subject to the periodic-grid correction above. Lines 16–17 sample the initial Gaussian function.
Lines 11–14 construct the discrete Fourier transform frequency order and $K^{2} = K_{X}^{2} + K_{Y}^{2}$ . Negative frequencies occupy the second half of each array. This is the discrete version of replacing each spatial derivative by multiplication by $ik$ .
Line 20 computes the two-dimensional Fast Fourier transform. Lines 22–24 multiply every mode by $exp (- α K^{2} Δ t)$ at every step, implementing the modal solution derived above.
Line 26 performs the inverse discrete Fourier transform. Taking the real part removes tiny imaginary roundoff errors; conjugate symmetry of the exact transformed real data already guarantees a real result.
Lines 28–42 display the initial and final arrays. A common colour scale is needed to compare their peak temperatures; independent automatic colour scales can hide the decrease in amplitude.

For this linear homogeneous problem, the exponential update is exact for the retained modes, apart from roundoff error. It is not a forward Euler method step and does not require the explicit finite-difference diffusion stability restriction. Sampling, periodic modelling and spatial resolution still matter.

 Read the full article

Past exam of the year 2 of the earth sciences (geology) course of the University of Oxford / 2025 / Paper 3 / 8 / b / Solution 2026-10-25

 View more

The wave equation is second order in time. Even one normal mode obeys a second-order ordinary differential equation, so its initial displacement and initial velocity are independent data. Physically they specify stored elastic deformation and the initial motion.

The heat equation is first order in time; its initial temperature already fixes the instantaneous time derivative through the spatial Laplacian. It needs one initial field. In either case spatial boundary conditions remain necessary and are not substitutes for initial data. Classical fixed-end string data must satisfy the appropriate endpoint compatibility, including

f (0) = f (L) = 0

and, when endpoint velocities exist,

g (0) = g (L) = 0

 Read the full article

Past exam of the zoology course of the University of Cambridge / 2024 / Paper ZL1 / 2 / Solution 2026-10-20

 View more

Mammalian endothermy probably emerged through several interacting changes along the synapsid stem, and its precise onset remains disputed because the fossil proxies measure different aspects of physiology. Endothermy concerns substantial metabolic heat production; homeothermy concerns relatively constant body temperature. A large animal can have relatively stable temperature without a mammalian metabolic rate, while an endothermic animal can display heterothermy. These distinctions prevent a single temperature estimate or anatomical character from being treated as a complete diagnosis.

The physiological problem is a heat balance. In a deliberately simplified steady-state model:

M - W = G (T_{b} - T_{a}) + E,

(1)

where

M

is metabolic energy turnover,

W

external work,

G

effective thermal conductance of an animal,

T_{b}, T_{a}

body and ambient temperature, and

E

evaporative heat loss. Maintaining a warm body in a cool environment can involve increased metabolism, reduced conductance, behaviour, or a combination. For example,

G = 0.1 W K^{- 1}

and a

20 K

difference imply

2 W

of non-evaporative heat loss in this illustrative model. The numbers are not a reconstruction of a fossil animal.

Along the synapsid lineage, increased sustained activity would require coordinated oxygen uptake, transport and use, greater food processing and sufficient energy assimilation. More effective locomotor construction, ventilation and circulation could support greater aerobic capacity, while insulation would reduce the cost of retaining the resulting heat. Shivering thermogenesis and later forms of non-shivering thermogenesis can contribute in living mammals. Brown adipose tissue in many living placental mammals cannot simply be assigned to early therapsids; a modern mechanism need not have been the ancestral one. Similarly, an expanded secondary palate or an altered limb posture is compatible with increased performance but is not a direct measurement of resting metabolism.

The time sequence must be reconstructed from several lines of evidence. Early synapsids lack the complete modern mammalian package, whereas various therapsids and cynodonts show changes in growth, feeding and respiratory anatomy before the mammalian crown radiation. Rapidly deposited fibrolamellar bone supports rapid growth, but fast-growing ectotherms and the influence of size and environment prevent a unique inference of endothermy. Growth interruptions can reflect seasonality or resource shortage rather than a categorical physiological boundary. Nasal structures interpreted as respiratory nasal turbinates could reduce water and heat loss during high ventilation, but identifying their precise form and separating respiratory from olfactory roles in fossils is difficult.

Preserved hair supplies direct evidence of insulation in some later fossils. Castorocauda, a Middle Jurassic mammaliaform, preserves integument alongside an aquatic anatomical specialization. ^[ref] This establishes that a fur-bearing covering existed by that time, not when insulation first evolved or whether its original function was thermal. Sensory and protective functions can precede elaboration for heat conservation. Direct preservation is particularly valuable, but its late appearance in the sampled record is not proof of biological absence in older lineages.

Stable isotope analysis has suggested earlier elevated thermometabolism. Comparing oxygen-isotope compositions of therapsid apatite with those of coexisting tetrapods, one study inferred differences in physiology and proposed elevated thermometabolism in several Permo-Triassic therapsid groups. ^[ref] The comparison is stronger than analysing one isolated animal because coexisting taxa constrain aspects of the environment. Nevertheless, body water, drinking source, habitat, evaporation and diagenesis affect the isotopic signal. The proposed middle-to-late Permian origin is an interpretation using a physiological framework, not a thermometer reading of an observed living cynodont.

A contrasting biomechanical approach uses the temperature dependence of endolymph properties and the dimensions of the semicircular canals. Calibration against living animals and application to fossil inner ears supported a relatively abrupt transition around 233 million years ago in the Late Triassic, near the origin of mammaliamorphs. ^[ref] This offers a quantitative prediction from a different anatomical system. However, body size, activity, reconstruction and the adequacy of the living calibration remain relevant. A change in a temperature-sensitive inner-ear proxy need not coincide exactly with the first increment in metabolic heat production or the first insulating covering. The disagreement with earlier isotope-based interpretations should therefore be tested, rather than erased by choosing the preferred date.

There are several plausible selective explanations. The aerobic capacity hypothesis proposes that selection for sustained locomotor performance raised maximum oxygen consumption, with elevated resting metabolism and heat production linked to that change. Comparative measurements of resting and maximum oxygen use underpin this proposal. ^[ref] A major issue is the strength and evolutionary direction of the relationship: high resting expenditure is expensive, and an association among living species does not prove which change led historically. Selection for thermal independence, nocturnal activity or performance in fluctuating climates provides related benefits, but “the climate became colder” is not sufficient unless the energetic costs and available food are considered.

A parental care hypothesis instead emphasizes selection for controlling the developmental environment and supplying offspring. Improved provisioning endurance and more stable incubation or offspring temperatures could favour increased metabolic capacity and heat retention. ^[ref] This gives concrete reproductive benefits that might offset energetic costs, while remaining a hypothesis about historical selective circumstances. The evidence required includes relationships among care, thermal performance and reproductive success, not merely the fact that living mammals both care for young and are endothermic. Activity, parental investment, insulation and environmental variability could reinforce one another, so the explanations need not be mutually exclusive.

The most secure account is a mosaic physiological transition with increasingly effective energy supply, heat production and heat retention. A Late Triassic onset of a more fully mammalian condition is a specific supported hypothesis, while earlier therapsid evidence indicates that parts of the transition may be older. fossil anatomy and chemistry constrain when particular features existed; the selective reasons require functional and comparative tests and remain less directly observable.

 Read the full article