E. coli B, Pasadena, 1958
The Band That Kept Half
Follow heavy nitrogen through the original centrifuge record. One intermediate DNA band contradicts conservative copying, later bands contradict simple dispersive copying, and a heat-split band reveals the two intact strands that semiconservative replication predicts.
Nitrogen-15 is stable. This was a density label, not a radioactive tracer. The dark lines in the paper were ultraviolet absorption photographs inside an analytical ultracentrifuge cell, not coloured bands in a test tube.
Move through the strip.
The culture was first grown with heavy nitrogen, then moved into ordinary nitrogen. Each row below preserves the reported sample label and the band species visible in Figure 4. Density increases to the right.
A structural redraw of Figure 4a and its adjacent microdensitometer trace.
source recordThe scan itself is not embedded because a reuse licence for the page image could not be verified. This is not a pixel digitization. Peak locations and species presence follow the paper; trace height, crop, colour, contrast, and fixed width are free display choices. Fractional generation labels came from population-growth estimates, not synchronized molecular rounds.
Give each model the same parent.
Start with one heavy-heavy duplex. Advance only whole copying rounds here. The browser creates every daughter molecule, then groups molecules with the same fraction of heavy nitrogen into density species.
Fixed-width Gaussians make discrete species visible. Their centers and weights come from the selected copying rule.
live modelEvery daughter molecule
| species | x | weight |
|---|
At one generation, conservative copying is gone. But one halfway band still permits two stories. The halfway band alone does not finish the experiment.
Heat the halfway band.
A fine-grained dispersive mosaic can also average to one halfway density. Meselson and Stahl heated first-generation hybrid DNA. Operate the same transformation on the ideal molecule.
The live rule separates a first-generation heavy-light duplex into its two component strands.
heat testOne heavy strand paired with one light strand.
one band
Computed from strand counts.
0.000 g/cm3
0.015 g/cm3 is a published scale input.
1.000
Display encoding of apparent molecular weight.
After heating, the paper reports two new species in equal amounts, 0.015 g/cm3 apart, each with approximately half the initial apparent molecular weight. It gives no numerical uncertainty for "equal" or "approximately half." The curve width shown here encodes that published approximate ratio; it is not a fresh measurement from Figure 9.
The check
The page keeps source inputs, live consequences, uncertainty, and display choices in separate lanes. The current three headline computations are repeated here.
Current ideal model
conservative, n = 1
x = 0.000 at 0.500; x = 1.000 at 0.500
Generation 1 anchor
x = 0.50 ± 0.02
The ±0.02 is the paper's estimated labelling-position error, not a standard deviation.
Heated hybrid
not yet separated
Apply heat to compute strand areas and source-scaled separation.
Published constants, displayed rather than derived
- Source samples: 0, 0.3, 0.7, 1.0, 1.5, 1.9, 2.5, 3.0, 4.1 estimated generations. Sampling-generation times were 0.81 h and 0.85 h.
- The Figure 4 samples were centrifuged for 20 h at 44,770 rpm, about 140,000 times g. Fully heavy and unlabelled reference DNA were separated by 0.014 g/cm3.
- The hybrid position was 50 ± 2% of the heavy-to-light interval. The paper estimates about 2% error for the position procedure. It does not call this a standard deviation or attach it point by point to the photographed trace.
- The heat treatment was 100 C for 30 min. The two products were reported in equal amounts, 0.015 g/cm3 apart, each at approximately 0.5 of the intact hybrid's apparent molecular weight.
Uncertainties and limits
- The direct scope is E. coli B under the reported culture, extraction, heating, and centrifugation conditions. This experiment did not inspect every organism or replicon.
- Generation 2 contradicts one progressively lightening band, the simple textbook dispersive prediction used here. It does not exclude every imaginable heterogeneous fragmentary mechanism. The authors' stronger heat conclusion was that the dispersive scheme proposed by Delbrück was ruled out.
- Generation 0 is calibration only because all three rules predict the same heavy band. At generation 1, semiconservative and simple dispersive predictions coincide at x = 0.5. Their difference there is identically zero, so agreement cannot choose between them.
- The culture was exponentially growing, not synchronized. The source scrubber never feeds fractional generation labels into the whole-round model.
Free choices in this page
- x = (rho - rho_light) / (rho_heavy - rho_light) is a chosen normalization. It preserves species positions and does not divide away the heavy-to-light effect.
- Gaussian width 0.055 x, source-redraw amplitudes, vertical scale, colour, contrast, crop, and line thickness are display choices. The widths are fixed, not reader controls, so there is no zero-contrast or deliberately unresolved setting that could manufacture a pass.
- The heat view uses sigma_single = sqrt(2) times sigma_intact to encode the paper's approximate half apparent molecular weight through M proportional to 1 / sigma squared. The resulting 0.500 is therefore an encoding check, not independent evidence.
Run the independent check: node research/meselson-stahl/verify-meselson-stahl.mjs. It enumerates every daughter molecule through four rounds, executes the page's model functions, integrates the heat curves, and fails if any published anchor or relation target drifts.
Why density becomes a band
In a centrifuged caesium chloride solution, solution density increases along the centrifugal field. DNA collects where the solution reaches its buoyant density. Diffusion spreads one molecular species around that position, giving an equilibrium band that is Gaussian under the method's ideal assumptions.
The 1957 method paper relates the square of band width inversely to molecular weight. That is why the 1958 heat experiment could read both a change of density and an approximate halving of apparent molecular weight from the same kind of record.
What the first band can and cannot say
Conservative replication keeps the heavy parent duplex intact and creates a new light duplex. After one round it therefore predicts separate heavy and light species. The observed single halfway species contradicts that rule.
Semiconservative copying puts one old heavy strand and one new light strand in each daughter. Simple dispersive copying mixes old and new material through both daughters. Both ideal rules place every first-generation molecule at the same average density, so the first band cannot distinguish them.