Animal anatomy

Animal anatomy

Each procedural animal has a template skeleton, Hill-type muscles, sampled segment mass and inertia, and a kinematic gait. You can inspect it in game mode or in engineering display mode. The quantities carry SI unit types. These models are not validated for engineering, clinical or animal-care decisions.

The source grades say how far each reference value was checked against its source. model_evidence() returns DESIGN for the selected body template, calibration and generated muscles. A checked source value does not validate the sculpt, its tissue assignment or a different species.

Call it

from extensions.anatomy.mode import ENGINEERING_MODE
from extensions.animals.anatomy.engineering import (
    calibrate,
    calibrated_animal,
    calibrated_mass,
)
from extensions.animals.anatomy.muscles import species_muscles
from extensions.animals.anatomy.render import materials_in_mode, mesh_in_mode
from extensions.animals.anatomy.stance import standing_loads
from extensions.animals.build import create_animal
from extensions.animals.gait import walk_pose
from extensions.animals.options import Variant, animal_options
from extensions.animals.registry import HORSE

# The selected body parameters are an explicit template estimate.
var cal = calibrate(HORSE, Variant(-1), allow_estimates=True)
var base = create_animal(HORSE, animal_options(3))
var horse = calibrated_animal(base, cal)
var mass = calibrated_mass(base, cal)
var muscles = species_muscles(horse.rig, HORSE, mass.total().mass)
var loads = standing_loads(horse, mass, muscles)
var pose = walk_pose(horse.rig, 0.25)
var meshes = mesh_in_mode(horse, muscles, pose, ENGINEERING_MODE)
var materials = materials_in_mode(ENGINEERING_MODE)

examples/animal_anatomy.mojo renders a horse, a dog and a cheetah in both modes. It also prints their mass, center of mass, stride and standing joint loads. make out/animal_anatomy.png runs it.

Game mode and engineering mode

AnatomyMode names the two modes. It lives in extensions/anatomy/mode.mojo, so other subjects can use it.

	Game mode	Engineering mode
Meshes	One painted mesh	Skin, skeleton and muscles
Surface	Painted coat, fur normals, baked occlusion	Flat tissue colors, no noise, no baked light
Coat	Drawn	Left out of the skin and of the mass
Muscles	Bellies bulge in the sculpt	Each belly is an ellipsoid of the muscle's volume
Numbers	Visual template	Inspectable SI estimates; no validated engineering capability

mesh_in_mode meshes an animal in a mode. materials_in_mode gives the materials for each mesh. In engineering mode the skin is translucent, so the skeleton and the muscles show through it.

Evidence grades

Each literature value carries one of five grades. Primary tables are available for the muscle inputs discussed below. Reading those tables improves traceability, but it does not independently verify their values. The reference grades remain at most FROM_TEXT.

The code and this page give the cited table and page. A value read only in a search-result extract or a secondary source stays at FROM_TEXT at best. Sources that could not be read lists what is still to check.

SpeciesBody.mass_source and length_source keep the provenance of each reference size. Muscle share_source and fiber_source keep the reference inputs. The separate model_evidence() reports DESIGN. Selected representative sizes, cross-species ratios, attachment paths and individualized architecture are modeling choices. That holds whatever the grade of the numbers they start from.

Grade	Meaning
FROM_ABSTRACT	The number is in the source's own abstract.
FROM_TEXT	The number is in the source's body, or in a secondary source that quotes it.
CROSS_CHECKED	Reserved for a documented independent check. Reading the primary table and reproducing its arithmetic do not qualify. The current animal muscle tables do not use it.
UNVERIFIED	No source that we read shows the number.
DESIGN	A modeling choice or a unit conversion, not a measurement.

Evidence.is_measured is True for the first three grades. Cited holds a grade and a source key from the reference list below.

Calibration

The sculpts are visual models. Calibration maps a canonical individual to a selected reference size and mass. Matching a prescribed mass is normalization, not an independent test of accuracy.

The calibration records its species and reference morph. It refuses use on another species or morph and refuses a second calibration of the same individual. A morph value of -1 permits only the coat-only variants that share one reference kind. Every numeric boundary rejects invalid or nonfinite input.

1. calibrate builds a canonical individual: an adult male of the selected reference morph.
2. reference_length measures an authored analytic envelope landmark, independently of the occupancy grid. For a quadruped mammal, this is the shoulder height. For a bird or a fish, it is the total length. For a frog, a rabbit, a rat, a snake or a spider, it is the body length.
3. The length factor maps this DESIGN landmark to the selected reference length.
4. It samples the scaled sculpt, with the physical coat depth unchanged. This gives predicted_mass before normalization. Multiplying the unscaled mass by scale cubed is not equivalent when coat depth is fixed.
5. Calibration.density_factor is the selected reference mass over that sampled prediction. At the canonical resolution, the normalized canonical mass equals the target by construction.

calibrated_animal scales the sculpt, the rig and the cells. The coat still paints. calibrated_mass scales the densities by the density factor. Each segment keeps the share of mass that the geometry gives it. An individual keeps its own size relative to the canonical one, so a juvenile stays small and light.

Every current selected body template has DESIGN parameter evidence and requires calibrate(..., allow_estimates=True). A matched morph and measured source values do not bypass this gate. Unmatched morphs and unverified reference inputs also require opt-in. Opt-in permits inspection of a template estimate; it does not mark the result validated. Raw visual and game construction remains available without this calibration opt-in.

An unmatched morph gets its own length factor but no mass correction from a different reference kind. For example, the bear reference describes a black bear. A grizzly has Calibration.matched False. matched means only that the named reference kind agrees, not that anatomy or accuracy was validated.

Landmark convention

reference_length uses analytic ellipsoid supports and round-cone end spheres. Fins use their polygon vertices and thickness envelope. Lenses use a conservative local rectangle. Carvers and smooth blends do not change this authored envelope. The classification selects withers, body or total-length solids. It does not measure eroded flesh.

The landmark scales consistently with geometry. It is independent of voxel spacing. It is also a DESIGN convention, not proof that the visible surface or a biological landmark has the cited dimension. Voxel-sampled extents are reported separately and can disagree, especially for thin animals.

What calibration does not establish

The canonical male mass is an input to normalization. It cannot also serve as independent validation. Female and juvenile outputs retain authored geometric proportions. Their agreement with a reference mass does not establish a measured growth or sex-specific model. No accuracy claim or universal sampling tolerance is made here.

What the calibration predicts

The calibration uses the selected adult male template. An adult female, drawn on her own, then gives a mass from her authored geometry and the same density normalization. This table compares that result with the selected female reference mass. These reference values are DESIGN parameters; agreement is not validation.

The check calibrates each species with calibrate(id, Variant(-1), allow_estimates=True). It draws the female with animal_options(1, quality=CROWD, sex=FEMALE, age=ADULT) and the selected reference morph, and weighs her with calibrated_mass at its default. That default is 40 cells a reference length, the resolution calibrate measures at.

Species	Predicted female mass	Selected female reference mass	Ratio
Bear	61.1 kg	60 kg	1.02
Boar	52.6 kg	70 kg	0.75
Cat	3.27 kg	3.5 kg	0.93
Cheetah	41.1 kg	30 kg	1.37
Chicken	1.42 kg	1.705 kg	0.83
Cow	533 kg	680 kg	0.78
Crow	0.390 kg	0.43 kg	0.91
Deer	49.5 kg	50 kg	0.99
Dog	26.0 kg	27 kg	0.96
Eagle	5.13 kg	5.2 kg	0.99
Fish	0.827 kg	1.0 kg	0.83
Fox	5.01 kg	5.0 kg	1.00
Frog	0.387 kg	0.30 kg	1.29
Goat	51.2 kg	65 kg	0.79
Horse	458 kg	450 kg	1.02
Lion	132 kg	126 kg	1.05
Pig	233 kg	250 kg	0.93
Rabbit	1.68 kg	1.8 kg	0.94
Rat	0.200 kg	0.25 kg	0.80
Shark	1470 kg	1400 kg	1.05
Sheep	67.4 kg	80 kg	0.84
Snake	0.421 kg	0.5 kg	0.84
Spider	23.9 g	20 g	1.19
Wolf	44.3 kg	45 kg	0.98

Seven species fall within 5 % of the selected targets. Fish, frog, rabbit and snake use the same selected mass for both sexes. Their rows compare authored female geometry with that target. The selected cheetah female mass, 30 kg, is not a measured mean from Hudson2011a. Its Table 1 gives female masses of 29.5 and 45.5 kg and three male masses from 27.5 to 32.0 kg.

Mass and inertia

sample_mass samples the posed sculpt on a grid of cubic cells:

• A cell's share of flesh is clamp(1/2 - d / h, 0, 1), with d the signed distance of its center and h its side.
• Solids of hair, wool and feathers are left out. The furred surfaces are pulled in by the species' coat depth. The coat reaches at most 30 % of a solid's smallest radius, so a thin leg keeps its flesh.
• Each cell selects the nearest flesh part after that part's coat erosion. It then takes the selected solid's density. It is added to the bone that the solid rides.

extensions/anatomy/inertia.mojo sums the cells. Each cell is a cuboid and adds its own inertia, m w^2 / 12 per axis. The parallel-axis theorem moves the tensor to the center of mass. The humanoid's limb segments use the same tally. The original extensions.humanoid.skeleton.tissue and soft_tissue import paths re-export the shared types and factories, so existing humanoid callers retain type identity.

The sampler checks primitive indexes, finite geometry and bounded grid dimensions before integer conversion or worker writes. A cone has two radii; its unused third radius does not remove coat erosion. Public tally addition rejects negative mass and invalid cell geometry. The result checks finite SI output and a physically admissible central tensor.

Finite primitive inputs can still produce unrepresentable derived values. The sampler rejects nonfinite or zero grid counts before integer conversion. It also checks sampled field values and nearest-solid indexes at run time. If a worker cannot represent a sampled field value, sample_mass raises an error after all workers finish; it does not return a partial mass. These checks apply with debug assertions disabled.

BodyMass.bone gives one bone's mass, center of mass and inertia. BodyMass.total gives the whole body. BodyMass.reference returns voxel-sampled extents. Thin features can be unresolved; a missing or nonpositive extent raises an error. These extents are not used for calibration or dimensional validation. Use reference_length for the separately defined template landmark.

Centralization removes residuals within a component's Float64 raw-moment roundoff bound before checking the tensor. This can discard small positive or negative residuals. It does not prove that the true moment is zero. A body whose size is unresolved relative to its offset from the origin needs a better-centered frame.

Densities

A mammal's segment takes Dempster's density for the homologous human segment (Dempster 1955, as tabulated in Winter 2009, Table 4.1). Applying human values to a quadruped is a DESIGN choice.

Segment	Density (g/cm^3)
Head and neck	1.11
Thorax	0.92
Abdomen and pelvis	1.01
Upper arm	1.07
Forearm	1.13
Hand	1.16
Thigh	1.05
Leg and tail	1.09
Foot	1.10

The other body plans take one whole-body density.

Body plan	Density (g/cm^3)	Grade
Chicken, plucked	1.044	FROM_TEXT (Hamershock 1993, Table 3, p. 11)
Flying bird, plucked	0.968	DESIGN: the mean of the eleven wild species of Hamershock 1993, Table 3, p. 11, which range from 0.880 to 1.050
Fish with a swim bladder	1.00	FROM_TEXT (Lindsey 2010)
Frog	1.00	DESIGN: a near-water whole-body approximation
Shark	1.05	UNVERIFIED
Snake	1.057	DESIGN: 1.13 with 6.5 % of the body lung
Spider	1.05	DESIGN

Keratin is 1.30 g/cm^3 (McKittrick 2012). Antler 1.8, tooth 2.1 and the eye 1.01 g/cm^3 are UNVERIFIED.

Muscles

extensions/anatomy/muscle.mojo is the shared Hill-type muscle.

• MuscleArchitecture holds the belly mass, the optimal fiber length, the pennation angle, the tendon slack length, the specific tension and the density.
• pcsa is m / (rho L0). max_force is the specific tension times the PCSA.
• The force curves are Thelen's (2003), with the young-adult parameters. The passive shape factor is 5 (Appendix, p. 75). The passive strain at F0 is 0.6 (Table 1, p. 71). The tendon strain at F0 is 0.04 (p. 71). The code used a shape factor of 4 before the paper was read.
• isometric_equilibrium solves the fibers and the tendon at one length.
• activation_rate is Thelen's first-order activation.

The default specific tension is 0.3 MPa. The muscle density is 1.06 g/cm^3 (Mendez and Keys 1960).

animal_muscles puts the limb muscles of a body plan on an individual's rig. A path starts at an origin, can wrap over via points on any bone, and ends at an insertion. AnimalMuscle.moment_arms gives the arm about each crossed joint in any pose. Each arm uses the stretch of the path that spans its joint. A positive arm turns the distal bone backward: it extends the hip and flexes the knee.

Mammal hind limb

The greyhound's biceps femoris anchors the template masses. From six cadavers, the mean muscle mass is 485 g and the mean body mass is 31.8 kg. Their ratio is about 1.53 % a side (FROM_TEXT). This is a ratio of cohort means, not a paired individual measurement. The muscle is in Williams 2008a, Table 1, on page 364, and the dogs' mass is in its Methods, on page 362.

The other muscles keep the rat's ratios to the biceps femoris and the rat's pennation (Eng 2008, Table 1, FROM_TEXT). The table is on page 2339. Eng defines projected PCSA as PCSA = m cos a / (rho L) in Methods, equation (3), using a density of 1.056 g/cm^3. Applying this equation to rounded table means need not reproduce the mean of specimen-level PCSAs. Mass scales with body mass, nearly isometrically (Pollock 1994, FROM_ABSTRACT).

Eng2008 reports fiber lengths normalized to a 2.4 micrometer sarcomere length. These are not bone lengths. The template divides them by fixed 41.0 mm femur and 46.2 mm tibia references that Eng2008 does not report. Their source remains untraced.

The current rat rig instead authors 36 mm and 41 mm segments. The ratios below remain DESIGN choices because the denominator is untraced. fiber_source grades each ratio as DESIGN and names Eng2008 only for the fiber input. The rig supplies each individual's segment lengths.

Muscle	Mass over the biceps femoris	Fiber length over its segment	Pennation
Biceps femoris	1	0.83 of the femur	3.6°
Rectus femoris	0.354	0.28 of the femur	25.4°
Vastus lateralis	0.483	0.48 of the femur	10.0°
Gastrocnemius, both heads	0.704	0.34 of the tibia	14.1°
Soleus	0.050	0.43 of the tibia	3.9°
Tibialis anterior	0.248	0.35 of the tibia	12.8°

A muscle of two or three heads sums their masses. Its fiber length is the mass-weighted harmonic mean of theirs, so volume over fiber length is additive before rounding. Its pennation is the mass-weighted mean. This is a DESIGN reduction of separate reference rows. It does not preserve each head's force direction or propagate sample variation. The plan's rat pennation values are DESIGN proxies when applied to other species.

Mammal fore limb

The greyhound gives the fore limb (Williams 2008b, Table 1, p. 375, seven dogs of 31.4 kg, FROM_TEXT). The triceps brachii is its long head, because only that head crosses both the shoulder and the elbow. The fiber length is a share of the dogs' humerus, 19.75 cm, or radius, 22.75 cm (Table 3, p. 376, FROM_TEXT).

Muscle	Mass	Share of body mass	Fiber length	Pennation
Triceps brachii, long head	341 g	1.09 %	6.5 cm	31°
Biceps brachii	54.1 g	0.17 %	1.8 cm	41°
Supraspinatus	150 g	0.48 %	5.9 cm	18°
Superficial digital flexor	18.3 g	0.058 %	1.2 cm	41°

Species tables

species_muscles and species_specs select reference rows for each species template. animal_muscles and plan_muscles keep the plan's muscles. Every dog, including the German Shepherd body template, uses greyhound muscle references. Crow and eagle also use cross-species proxies. These transfers remain DESIGN.

A row can give an absolute fiber length, scaled by the cube root of body mass relative to a selected reference mass. Payne2005, Table 5, uses geometric similarity for comparisons; it does not validate this template. The direct MuscleSpec.fiber_length helper rejects invalid inputs and nonpositive or nonfinite results.

Species	Source	Muscles	Grade
Dog	Williams 2008a, Table 1, p. 364 (31.8 kg greyhounds). The soleus is from Hudson 2011a, Table 3, p. 367 (greyhounds of 27.3 kg).	Hind limb	FROM_TEXT
Cheetah	Hudson 2011a, Table 3, p. 367, and 2011b, Table 2, p. 378. The body mass, 33.1 kg, is the mean of the five weighed cheetahs (Table 1).	Hind and fore limb	FROM_TEXT: the means of ratios do not give the mean PCSA. Pennation uses greyhound proxies, DESIGN, except for the soleus, which retains the rat plan's 3.9° as a DESIGN proxy. The 33.1 kg normalization uses five known body masses, while most muscle rows include eight subjects.
Horse	Payne 2005, Table 4, p. 561. The body mass, 510 kg, is the mean of all seven subjects in Table 3: five Thoroughbreds, one Thoroughbred cross and one Arab.	Hind limb	FROM_TEXT
Rat	Eng 2008, Table 1, p. 2339. The body mass, 323 g, is on p. 2337.	Hind limb	FROM_TEXT
Chicken	Hartman 1961, Table 1, p. 45	Flight muscles	FROM_TEXT
Crow	Hartman 1961, Table 1, p. 71	Flight muscles	FROM_TEXT
Eagle	Hartman 1961, Table 1, p. 43	Flight muscles	FROM_TEXT

Other muscles

• Mammal fore limb: the superficial digital flexor's tendon runs behind the carpus and over the sesamoids, so it holds the wrist and the fetlock.
• Bird: the template uses the medians of all 42 rows of Hartman1961, Table 3, p. 89. The bilateral pectoralis median is (14.75 + 15.00) / 2 = 14.875 %. The supracoracoideus median is (1.39 + 1.44) / 2 = 1.415 %. Halving gives 7.4375 % and 0.7075 % a side. These summaries are DESIGN choices from FROM_TEXT inputs. The species rows come from Hartman's Table 1. The White Leghorn's pectoral muscles are 10.6 % of body mass (20 birds, p. 45), split 8.78 to 3.50 as in the one bird weighed by muscle. The American crow's are 14.2 % (Corvus brachyrhynchos pascuus, 3 birds, p. 71), split as in Cyanocorax affinis on the same page. Hartman has no golden eagle, so the eagle takes the mean of five accipitrids on p. 43: a pectoralis of 13.4 % and a supracoracoideus of 0.49 %. The fiber lengths are DESIGN shares of the humerus. The supracoracoideus lifts the wing over its pulley, the triosseal canal. The wing beats about the body's long axis.
• Frog: the hind limbs hold 33 % of body mass (James 2008, abstract, FROM_ABSTRACT for that number, Litoria nasuta). The plantaris is 19.4 % of that (FROM_TEXT). The cruralis and the semimembranosus are DESIGN shares.
• Fish, shark and snake: axial_muscles gives each spine joint a muscle a side. Its mass is a share of its segment's sampled mass. The share is 60 % for a fish (UNVERIFIED) and 53 % for a shark (Bernal 2003 for the red muscle). It is 30 % for a snake (DESIGN).
• Spider: each femur-patella and tibia-metatarsus joint has one flexor. A spider has no extensor at those joints. Hemolymph pressure extends them: hemolymph_pressure gives 6.5 kPa at rest and 65 kPa at a peak (Parry 1959; Anderson 1975). The flexors develop 0.5 MPa (Medler 2002).

The attachment offsets are DESIGN choices that give each muscle its known action. The tendon slack length makes the fibers optimal in the standing bind pose.

Standing loads

standing_loads gives the static load on each left limb joint of a standing quadruped:

1. Each foot presses at the middle of its sole.
2. The center of mass splits the weight between the fore and the hind feet.
3. About each joint, the foot's push and every distal segment's downward weight have moments. The muscles must cancel their sum.
4. The muscles that oppose the net moment share one fiber stress. Their tendon force includes the cosine of pennation. The least specific tension among the participating muscles limits the shared stress. Muscle ordering does not change the result.

JointLoad gives the moment, the ground lever, the muscle lever, the effective mechanical advantage (Biewener 1989), the stress and the activation. A nonzero demand that its muscles cannot hold at full activation reports held False. Zero required moment reports held True with zero activation. This does not prove that tendons, ligaments or any other passive structure can supply the missing moment. Those structures are not modeled.

This is a symmetric sagittal-plane estimate. It does not solve lateral balance, contact pressure, tendon compliance or three-dimensional stability.

The moment arms come from DESIGN attachments that scale with the segments. So the effective mechanical advantage does not grow with body mass as Biewener's M^0.26 does.

Gait timing

extensions/anatomy/locomotion.mojo holds dynamic similarity. Animals move alike at equal Froude numbers, Fr = u^2 / (g h) (Alexander and Jayes 1983). The stride length is 2.3 Fr^0.3 h (Alexander 1976, FROM_TEXT).

walk_pose takes a Froude number. The default 0.25 is a design choice. Reference stride, frequency and speed are dynamic-similarity estimates, not a validated locomotion model.

walk_stride caps the visual stride at the shortest reachable two-link stance sweep. The pose compensates root bob at planted feet. The animal walks in place. Its equivalent translation speed is visual stride times reference frequency.

walk_pose_at takes a Duration, so different animals advance on one physical timeline at their own reference frequencies. The comparison gallery plays once because their periods differ.

Each foot sweeps 70 % of the stride under the body. This duty factor is a DESIGN value: Alexander and Jayes 1983 give duty factor against Froude number, but no copy of the paper was readable.

Bulging bellies

The engineering muscle belly uses its assigned mass and density to define an ellipsoid volume. The game sculpt uses an approximate visual bulge. flexed reshapes the tagged solids for a pose.

When a muscle's fibers shorten by a factor f, its solids shrink by f^w along the muscle and grow by f^(-w/2) across it. w is 0.5, a DESIGN value: the skin and the fat do not bulge. f is held to 0.6 to 1.4. The bind pose changes no solid.

Ellipsoid factors have product one. Cone radii change while endpoints stay fixed, so cones and smooth unions do not conserve tissue volume.

flexed_animal returns a marked visual-only individual with the reshaped sculpt. Mesh it with the same pose. sample_mass and calibrated_mass refuse this visual-only geometry. Sample the original calibrated anatomy instead.

Skeleton

skeleton_model draws a round cone along each bone. Its long-bone radius ratios use mid-shaft diameters and bone lengths from five cheetahs and three greyhounds. These inputs are FROM_TEXT (Hudson2011a, Table 1, and Hudson2011b, Table 1). Averaging these ratios and applying them to every mammal is DESIGN.

Bone	Radius over length
Femur	0.0385
Tibia	0.036
Humerus	0.0456
Radius	0.0337

The model uses one ratio at every size as a DESIGN simplification. The source exponents for length and diameter, M^0.35 and M^0.36 (Alexander1979, FROM_ABSTRACT), differ and do not establish an exact constant ratio. The other bones' ratios are also DESIGN.

Physics bodies

segment_bodies turns each bone with flesh into a dynamic RigidBody for extensions/physics. Each body has the sampled mass, center of mass and inertia tensor. to_physics turns the sculpt's +y-up frame into the physics world's +z-up frame. The physics world has no joints yet, so the bodies are the segments of a multibody model.

Species

All selected masses and reference lengths below are DESIGN parameters. The final column records only the grade of the cited source.

Species	Kind	Selected male, female mass	Selected reference length	Source grade
Bear	American black bear	100, 60 kg	0.90 m shoulder	FROM_TEXT
Boar	Central European wild boar	85, 70 kg	0.75 m shoulder	FROM_TEXT
Cat	Domestic cat	4.5, 3.5 kg	0.25 m shoulder	FROM_TEXT
Cheetah	Cheetah	50, 30 kg	0.80 m shoulder	FROM_TEXT
Chicken	White Leghorn	2.43, 1.705 kg	0.45 m total	mass FROM_TEXT (Hartman 1961, Table 1, p. 45), length UNVERIFIED
Cow	Holstein-Friesian	900, 680 kg	1.47 m shoulder	mass UNVERIFIED: Holstein USA gives the cow, 1500 lb, and only an upper bound for the bull. The height is the cow's.
Crow	American crow	0.47, 0.43 kg	0.45 m total	FROM_TEXT
Deer	White-tailed deer	70, 50 kg	0.90 m shoulder	FROM_TEXT
Dog	German Shepherd Dog	35, 27 kg	0.625 m shoulder	FROM_TEXT: the middles of FCI-Standard No. 166, p. 8
Eagle	Golden eagle	3.7, 5.2 kg	0.85 m total	FROM_TEXT
Fish	Rainbow trout	1.0 kg	0.45 m total	FROM_TEXT
Fox	Red fox	6.5, 5.0 kg	0.40 m shoulder	FROM_TEXT
Frog	American bullfrog	0.30 kg	0.155 m snout-vent	mass UNVERIFIED: the sources found give only an upper bound, 0.5 kg
Goat	Saanen dairy goat	85, 65 kg	0.94 m shoulder, the buck's	mass UNVERIFIED: NSW DPI gives only the doe's minimum, 64 kg
Horse	Thoroughbred	500, 450 kg	1.62 m shoulder	FROM_TEXT. Payne 2005, Table 3, p. 561, lists seven horses: five Thoroughbreds, one Thoroughbred cross and one Arab. The Thoroughbreds weigh 480 to 600 kg and stand 1.47 to 1.57 m. The full cohort has a mean mass of 510 kg. Sex is not given.
Lion	African lion	190, 126 kg	1.15 m shoulder	FROM_TEXT
Pig	Large White	300, 250 kg	0.90 m shoulder	length UNVERIFIED. 2068 Large White sows of 100 kg stand 0.614 m (Hong 2021, Table 1). Isometric scaling to 300 kg gives 0.886 m.
Rabbit	European rabbit	1.8 kg	0.38 m head-body	FROM_TEXT
Rat	Norway rat	0.30, 0.25 kg	0.21 m head-body	FROM_TEXT
Shark	White shark	850, 1400 kg	3.7 m total	FROM_TEXT
Sheep	Suffolk	130, 80 kg	0.70 m shoulder	FROM_TEXT
Snake	Corn snake	0.5 kg	0.87 m snout-vent	FROM_TEXT
Spider	Mexican redknee tarantula	12, 20 g	0.055 m body	FROM_TEXT
Wolf	Gray wolf	55, 45 kg	0.80 m shoulder	FROM_TEXT

Limits

• The muscle paths are straight lines between attachments and via points. They do not wrap around bone surfaces.
• The tendons are rigid when bellies bulge. The static solve, isometric_equilibrium, has an elastic tendon.
• The individual muscle architecture, selected representative body parameters, attachments and coat depths are DESIGN values. Source evidence is recorded separately.
• The cheetah uses greyhound pennation proxies except for soleus. Dog and cheetah soleus retain the rat plan's 3.9° as a DESIGN proxy because their reference rows give no angle. The horse and rat use the plan's forelimb references.
• Birds have flight muscles but no leg muscles.
• The physics world has no joints, so the segment bodies do not yet form a chain
• A normalized total mass does not validate segment fractions, centers, tensors, tissue densities or muscle capacity
• The engineering display grants no validated engineering capability

Types and checks

Each id, kind and mode is a type with is_valid: AnatomyMode, Evidence, BodyPlan, ReferenceKind, BodyTissue and Segment. Every boundary that reads one refuses a value that is not valid. The files in tests/compile_fail/ prove that a bare integer does not compile.

Sources that could not be read

These sources are behind a paywall, and no open copy was found. The values that rely on them keep their grade.

• Alexander 1979 and 1981, and Alexander and Jayes 1983, in J. Zool. The bone exponents come from the 1979 abstract. The duty factor stays DESIGN.
• Sacks and Roy 1982 (cat hind limb) and Lieber and Blevins 1989 (rabbit hind limb), in J. Morphol. The cat and the rabbit keep the plan's muscles.
• James and Wilson 2008, in Physiol. Biochem. Zool. Only the abstract was read.

Paxton 2010 was read. It measures broiler and junglefowl hind limbs, and the bird plan has no leg muscles, so the code does not use it.

References

• Alexander1976: Alexander, R. McN. Estimates of speeds of dinosaurs. Nature 261:129-130, 1976. The stride relation was read as a quotation in later trackway papers.
• Alexander1979: Alexander, Jayes, Maloiy and Wathuta. Allometry of the limb bones of mammals from shrews (Sorex) to elephant (Loxodonta). J. Zool. 189:305-314, 1979. Only the abstract was read.
• Alexander1981: Alexander, Jayes, Maloiy and Wathuta. Allometry of the leg muscles of mammals. J. Zool. 194:539-552, 1981. Not read.
• AlexanderJayes1983: Alexander and Jayes. A dynamic similarity hypothesis for the gaits of quadrupedal mammals. J. Zool. 201:135-152, 1983. Not read.
• Anderson1975: Anderson and Prestwich. The fluid pressure pumps of spiders (Chelicerata, Araneae). Z. Morph. Tiere 81:257-277, 1975.
• Bernal2003: Bernal, Sepulveda, Mathieu-Costello and Graham. Comparative studies of high performance swimming in sharks I. Red muscle morphometrics, vascularization and ultrastructure. J. Exp. Biol. 206:2831-2843, 2003.
• Biewener1989: Biewener. Scaling body support in mammals: limb posture and muscle mechanics. Science 245:45-48, 1989.
• Dempster1955: Dempster. Space requirements of the seated operator. WADC Technical Report 55-159, 1955.
• Eng2008: Eng, Smallwood, Rainiero, Lahey, Ward and Lieber. Scaling of muscle architecture and fiber types in the rat hindlimb. J. Exp. Biol. 211:2336-2345, 2008. doi:10.1242/jeb.017640. Primary text. Primary source locations: Table 1, p. 2339; the rats' mass, p. 2337.
• FCI166: Fédération Cynologique Internationale. FCI-Standard No. 166, German Shepherd Dog. Official PDF. Primary source locations: size and weight, p. 8.
• Hamershock1993: Hamershock, Seamans and Bernhardt. Determination of body density for twelve bird species. Technical Report WL-TR-93-3049, Wright Laboratory, 1993. DTIC ADA266452. Primary source locations: Table 3, p. 11.
• Hartman1961: Hartman. Locomotor mechanisms of birds. Smithsonian Misc. Coll. 143(1):1-91, 1961. Primary PDF. Primary source locations: Table 1, pp. 43, 45 and 71; Table 3, p. 89; Methods, p. 2.
• Holstein: Holstein Association USA, breed history page. Read only as a search-result extract.
• Hudson2011a: Hudson, Corr, Payne-Davis, Clancy, Lane and Wilson. Functional anatomy of the cheetah (Acinonyx jubatus) hindlimb. J. Anat. 218:363-374, 2011. doi:10.1111/j.1469-7580.2010.01310.x, PMC3077520. Primary source locations: Table 1, p. 364; Table 3, p. 367.
• Hudson2011b: Hudson, Corr, Payne-Davis, Clancy, Lane and Wilson. Functional anatomy of the cheetah (Acinonyx jubatus) forelimb. J. Anat. 218:375-385, 2011. doi:10.1111/j.1469-7580.2011.01344.x, PMC3077521. Primary source locations: Table 1, p. 377; Table 2, p. 378.
• James2008: James and Wilson. Explosive jumping: extreme morphological and physiological specializations of Australian rocket frogs (Litoria nasuta). Physiol. Biochem. Zool. 81:176-185, 2008. PubMed 18190283. Only the abstract was read.
• Lindsey2010: Lindsey, Smith and Croll. From inflation to flotation: contribution of the swimbladder to whole-body density and swimming depth during development of the zebrafish (Danio rerio). Zebrafish 7:85-96, 2010.
• McKittrick2012: McKittrick, Chen, Bodde, Yang, Novitskaya and Meyers. The structure, functions, and mechanical properties of keratin. JOM 64:449-468, 2012.
• Medler2002: Medler. Comparative trends in shortening velocity and force production in skeletal muscles. Am. J. Physiol. Regul. Integr. Comp. Physiol. 283:R368-R378, 2002.
• MendezKeys1960: Mendez and Keys. Density and composition of mammalian muscle. Metabolism 9:184-188, 1960.
• NSWDPI: NSW Department of Primary Industries. Goat breeds: Saanen. Read only as a search-result extract.
• Parry1959: Parry and Brown. The hydraulic mechanism of the spider leg. J. Exp. Biol. 36:423-433, 1959.
• Paxton2010: Paxton, Anthony, Corr and Hutchinson. The effects of selective breeding on the architectural properties of the pelvic limb in broiler chickens: a comparative study across modern and ancestral populations. J. Anat. 217:153, 2010. PMC2913024. Read, not used.
• Payne2005: Payne, Hutchinson, Robilliard, Smith and Wilson. The pelvic limb anatomy of horses (Equus caballus). J. Anat. 206:557-574, 2005. PMC1571521. Primary source locations: Tables 3 and 4, p. 561.
• Pollock1994: Pollock and Shadwick. Allometry of muscle, tendon, and elastic energy storage capacity in mammals. Am. J. Physiol. 266:R1022-R1031, 1994.
• Thelen2003: Thelen. Adjustment of muscle mechanics model parameters to simulate dynamic contractions in older adults. J. Biomech. Eng. 125:70-77, 2003. doi:10.1115/1.1531112. Primary source locations: Table 1, p. 71; Appendix, p. 75.
• Hong2021: Hong, Ye, Dong, Li, Yan, Cai, Liu, Tan and Wu. Genome-wide association study for body length, body height, and total teat number in Large White pigs. Front. Genet. 12:650370, 2021. PMC8366400. Primary source locations: Table 1.
• Williams2008a: Williams, Wilson, Rhodes, Andrews and Payne. Functional anatomy and muscle moment arms of the pelvic limb of an elite sprinting athlete: the racing greyhound (Canis familiaris). J. Anat. 213:361-372, 2008. PMC2644771. Primary source locations: Methods, p. 362; Table 1, p. 364; Table 3, p. 366.
• Williams2008b: Williams, Wilson, Daynes, Peckham and Payne. Functional anatomy and muscle moment arms of the thoracic limb of an elite sprinting athlete: the racing greyhound (Canis familiaris). J. Anat. 213:373-382, 2008. PMC2644772. Primary source locations: Methods, p. 374; Table 1, p. 375; Table 3, p. 376.
• Winter2009: Winter. Biomechanics and Motor Control of Human Movement, 4th ed., Wiley, 2009. The densities were read from two code copies of Table 4.1 that agree.

The other species masses and lengths come from field guides and breed sources. These include the Bear Specialist Group, Animal Diversity Web, the San Diego Zoo Wildlife Alliance and the Florida Museum. They were read as search-result extracts, so their grade is FROM_TEXT at best. Their keys in extensions/animals/anatomy/body.mojo name them.