Step 1: multimeter

Set the meter to resistance and touch the probes to two clean points about a centimetre apart. Quartz, tourmaline, glass and almost all stones read “OL”, meaning an open circuit: they are insulators. Iron meteorites, like all metals, read a fraction of an ohm. A “crystal” that conducts is not quartz; a “meteorite” that does not conduct is not an iron meteorite (though it may still be a stony one).

Step 2: magnet

A strong neodymium magnet clings firmly to iron meteorites, and weakly to most stony meteorites, which contain grains of metal. But many terrestrial materials are magnetic too: steel, industrial slag, magnetite, and the shiny black beads sold as “magnetic hematite”, which are a manufactured ferrite. Attraction is a necessary condition for an iron meteorite, not a sufficient one.

Step 3: loupe on an etched face

A Widmanstätten pattern appears only on a polished face that has been etched. Under a loupe, a genuine octahedrite shows kamacite bands of consistent width crossing in three or four directions, with thin bright taenite borders and fine-grained plessite between them; a hexahedrite shows only fine straight Neumann lines; an ataxite shows nothing. Only laboratory analysis of nickel content (typically by X-ray fluorescence or electron microprobe) confirms a meteorite, and named meteorites are registered in the Meteoritical Society’s database.

Step 4: irradiance meter

Measure the irradiance of a red-light device at the distance where the skin will actually be. Use a meter suited to narrow-band light, such as a thermopile power meter with a known aperture or a spectroradiometer; meters designed for sunlight can misread LEDs badly.

def stone_check(ohms_1cm, magnet, etched_pattern):
    """Steps 1–3: multimeter across ~1 cm, a strong magnet, a 10× loupe on an etched face.
    ohms_1cm: None for 'OL' (open). magnet: 'strong' | 'weak' | 'none'.
    etched_pattern: 'crossing bands' | 'fine lines' | 'none'."""
    if ohms_1cm is None:
        return "insulator: quartz, tourmaline, glass, stone; not an iron meteorite"
    if magnet != "strong":
        return "conductor but not strongly magnetic: not iron-nickel metal"
    if etched_pattern == "crossing bands":
        return "consistent with an octahedrite; confirm nickel in a laboratory"
    if etched_pattern == "fine lines":
        return "consistent with a hexahedrite (Neumann lines); confirm Ni in a laboratory"
    return "iron, steel, magnetite or a rare ataxite: needs a laboratory"

for sample in ((None, "none", "none"), (0.2, "strong", "crossing bands"),
               (0.3, "strong", "none"), (2.0, "none", "none")):
    print(f"{str(sample):40s} → {stone_check(*sample)}")

# Step 4: a red-light panel. Label claims 100 mW/cm²; readings at the skin position:
label = 100.0
measured = {5: 62.0, 15: 35.0, 30: 18.0}           # cm → mW/cm², example readings
print()
for cm, mw in measured.items():
    print(f"{cm:2d} cm: {mw:5.1f} mW/cm² ({mw / label:.0%} of label); "
          f"4 J/cm² takes {4 / (mw / 1000):4.0f} s")
# (None, 'none', 'none')                   → insulator: quartz, tourmaline, glass, stone; not an iron meteorite
# (0.2, 'strong', 'crossing bands')        → consistent with an octahedrite; confirm nickel in a laboratory
# (0.3, 'strong', 'none')                  → iron, steel, magnetite or a rare ataxite: needs a laboratory
# (2.0, 'none', 'none')                    → conductor but not strongly magnetic: not iron-nickel metal
#
#  5 cm:  62.0 mW/cm² (62% of label); 4 J/cm² takes   65 s
# 15 cm:  35.0 mW/cm² (35% of label); 4 J/cm² takes  114 s
# 30 cm:  18.0 mW/cm² (18% of label); 4 J/cm² takes  222 s

The example readings, which are invented, show a pattern that is common in practice: the irradiance at a comfortable distance is a fraction of the figure on the label, often measured at the surface of the lens. A user who follows the label’s suggested time at 30 cm receives perhaps a fifth of the intended dose; one who moves closer to compensate also receives more heat.

Question

Does a short exposure to red and near-infrared light increase blood flow in healthy forearm skin more than a sham that produces the same warming? Chapter 5 explained why this is the right comparison: warmth alone increases skin perfusion, so a study that compares light with darkness cannot separate a photochemical effect from heat.

Design

A variant of the protocol for muscle-fatigue recovery keeps the same devices, sham and blinding, applies the light over the forearm flexors after a standard fatiguing handgrip task, and takes as its primary outcome the recovery of maximal grip force 20 minutes later; its sample size is calculated in the same way from pilot data.

Sample size

import math
from statistics import NormalDist

z = NormalDist().inv_cdf

def completers(sd_diff, delta, alpha=0.05, power=0.90):
    """Two-period crossover analysed as paired differences."""
    za, zb = z(1 - alpha / 2), z(power)
    return math.ceil(((za + zb) * sd_diff / delta) ** 2 + za ** 2 / 2)

# Outcome: change in forearm skin perfusion (laser speckle contrast imaging),
# % of the pre-exposure baseline, 10 minutes after exposure ends.
SESOI = 10.0                                    # percentage points
for sd in (15.0, 20.0, 25.0):
    n = completers(sd, SESOI)
    print(f"sd of within-person difference {sd:.0f} points: {n} completers "
          f"(recruit {math.ceil(n / 0.8)} if 20 % drop out)")
# sd of within-person difference 15 points: 26 completers (recruit 33 if 20 % drop out)
# sd of within-person difference 20 points: 44 completers (recruit 55 if 20 % drop out)
# sd of within-person difference 25 points: 68 completers (recruit 85 if 20 % drop out)

The protocol plans for the middle case: 44 participants completing both sessions, 55 recruited to allow for drop-out. An eight-person pilot checks the heat matching and the procedures only.

Analysis and decision rule

The primary analysis estimates the within-person difference in perfusion change (active minus sham) with its 95 % confidence interval, with period and sequence in the model to check for carry-over. The result will be interpreted by the following rule, written before any data exist, in the spirit of equivalence testing (Lakens 2017).

SESOI = 10.0     # percentage points of perfusion change: smallest difference worth acting on

def verdict(diff, lo, hi):
    """diff = active light minus heat-matched sham, in perfusion change
    (percentage points of baseline), with its 95 % confidence interval.
    Fixed before any data are seen."""
    if lo >= SESOI:
        return "active light clearly better (by at least 10 points)"
    if lo > 0:
        return "active light better; size not shown to reach 10 points"
    if hi < 0:
        return "sham better"
    if lo >= -SESOI and hi <= SESOI:
        return "no difference worth acting on (within ±10 points)"
    return "unclear: the interval is too wide"

for d, lo, hi in ((16, 11, 21), (6, 1, 11), (2, -4, 8), (5, -6, 16)):
    print(f"{d:+3d} ({lo:+3d} to {hi:+3d}): {verdict(d, lo, hi)}")

# Blinding check: share of participants who guess their condition correctly.
guesses_right, n = 25, 44
print(f"correct guesses {guesses_right}/{n} = {guesses_right / n:.0%} (50 % = chance)")
# +16 (+11 to +21): active light clearly better (by at least 10 points)
#  +6 ( +1 to +11): active light better; size not shown to reach 10 points
#  +2 ( -4 to  +8): no difference worth acting on (within ±10 points)
#  +5 ( -6 to +16): unclear: the interval is too wide
# correct guesses 25/44 = 57% (50 % = chance)

The examples are invented, to show in advance how each kind of result will be reported. An interval within ±10 points will be reported as “no difference worth acting on”, not “no effect”; an interval that crosses a boundary will be reported as unclear, not as a “trend”. If participants guess their condition much better than chance, the report will say that blinding failed, and the result will be treated as weaker evidence.

Ethics and conflicts of interest

Both conditions are brief, low-intensity exposures of the forearm. Participants give informed consent, may leave at any time, and are screened by the criteria above. The protocol will be reviewed by an independent research ethics committee and publicly registered before recruitment, following the principles of pre-registration (Nosek et al. 2018).

What this experiment cannot show

It tests one dose, one pair of wavelengths and one outcome, in the healthy skin of the forearm, over half an hour. It says nothing about any disease, about the brain, about long-term use, or about other doses. A positive result would show that red and near-infrared light do something beyond warmth in healthy skin; it would not show that any device treats anything.