Astronomy · The matter between stars

Tracing star formation with several observations

Compare complementary formation indicators and reproduce two invented finite-window birth-rate summaries that respond differently to the same total formed mass.

Continue in the approved chapter order. Necessary vocabulary, quantities and assumptions are explained in the lesson.

Back to the chapter

After reading, you can

Distinguish formation-rate inference from a direct instantaneous measurement.
Calculate short- and long-window average birth rates with declared units.
Explain how response times, population accounting and contamination affect comparisons among tracers.

A formation indicator has a physical response

Star formation can be investigated through identified young stars and through radiation associated with young populations and their surroundings. Ultraviolet emission, ionized-gas lines, infrared dust emission and radio measurements supply complementary information. They are not interchangeable direct counters of stars being born at one exact instant.

Counting young stars requires an age or lifetime interpretation and a correction for objects the observations could not identify. Turning a count into total stellar birth mass also requires the distribution of initial masses and a convention for stars versus unresolved systems.

Ionized-gas emission responds to the supply and fate of ionizing radiation. Ultraviolet measurements can weight a different stellar population and range of ages. Dust emission responds to absorbed radiation from its heating sources, while radio indicators have their own emission mechanisms and calibration assumptions.

The review by Kennicutt and Evans connects such diagnostics to population models, attenuation and calibration. The exact response depends on the chosen observable and circumstances. No universal coefficient or actual tracer response timescale is assigned in the following original exercise.

Define two finite-window summaries

At one analysis epoch, let age τ be lookback time since a parcel of stellar birth mass formed: τ = 0 is the present and larger τ is further in the past. Let t0 be an arbitrary positive time unit and M0 an arbitrary positive mass unit.

Supply birth mass M_recent formed between lookback ages 0 and 2t0, and M_older formed between 2t0 and 10t0. Birth mass means the mass initially put into the stellar population, rather than its current mass after material return. No formation outside this finite age range contributes to either summary.

Define invented short-window average A = M_recent/(2t0) and long-window average B = (M_recent + M_older)/(10t0), in mass per time. Equivalently, weight the birth-rate history by a rectangular response of height 1/(2t0) over 0–2t0 or 1/(10t0) over 0–10t0. Each response has unit integral.

These are mathematical summaries of supplied birth mass. They are not measured photon signals or actual H-alpha, ultraviolet or infrared response functions. The history inside either bin is unspecified, so neither summary supplies the instantaneous rate at the present endpoint.

Equal total birth mass can hide different recent activity

History I has recent and older birth masses (6, 24)M0. The short average is 6/2 = 3M0/t0 and the long average is 30/10 = 3M0/t0. Its two bin-average rates agree, but the unspecified history need not have been perfectly constant at every time.

History II has (2, 28)M0, giving short average 1M0/t0 and the same long average 3M0/t0. History III has (10, 20)M0, giving short average 5M0/t0 and again long average 3M0/t0. All three histories formed the same 30M0 across the full interval.

Their A/B ratios are 1, 1/3 and 5/3. A discrepancy between summaries can therefore arise from their different time weighting even when the supplied birth masses and arithmetic are exact. It need not mean that a measurement failed.

The table and schematic show only the two declared finite windows and these integrated masses. The vertical response height carries inverse-time units; the integral of response times a birth rate has rate units. A unit-area response is not a probability that a particular star formed.

Original finite-window birth-mass histories in arbitrary M0 and t0. A averages 0–2t0 and B averages 0–10t0, using invented unit-area rectangular responses. These are mathematical summaries, not real tracer calibrations or instantaneous rates.
History	M_recent/M0	M_older/M0	A/(M0/t0)	B/(M0/t0)	A/B
I: equal bin-average rates	6	24	3	3	1
II: lower recent average	2	28	1	3	1/3
III: higher recent average	10	20	5	3	5/3

Different observations test different parts of a model

Real indicators have smoother and more complicated responses than these rectangles. Stellar lifetimes, composition, initial mass distribution and the formation history affect the luminosity produced per unit birth mass. Rare luminous stars can also be incompletely sampled in a small population.

Absorption and escape of radiation can alter the relation between stellar production and observed gas emission. Dust attenuation changes received ultraviolet and optical light. Infrared emission can miss escaped starlight or include heating by older stars and other sources.

Two estimates should therefore be compared for a consistent aperture, distance treatment, bands, population accounting and uncertainty model. A small region and a whole galaxy need not satisfy the same population-sampling or heating assumptions.

Agreement is useful when the methods constrain different failure modes, but shared assumptions can also produce shared systematic errors. Disagreement by itself does not uniquely determine a dust correction, an initial mass distribution or a recent formation history.

Two windows, two averages

Scroll sideways to see the full diagram. Its explanation is below.

Original invented response-window model, not observed data, H-alpha or ultraviolet response curves, or actual tracer timescales. The horizontal coordinate is lookback age tau divided by a positive arbitrary time unit t0: zero is the analysis epoch and larger values to the right are older birth times. Each kernel K has inverse-time units. The vertical coordinate t0 K is dimensionless. Both panels use the same linear horizontal range from 0 to 10 and vertical range from 0 to 0.5, with explicit zeros. Kernel A equals 1/(2 t0) for lookback ages from 0 to 2 t0 and is zero outside this finite interval. Kernel B equals 1/(10 t0) from 0 to 10 t0 and is zero outside that finite interval. A is a tall narrow blue rectangle with solid outline and B is a short wide orange rectangle with dashed outline. The boundaries are exact steps, not smoothed curves. Endpoint choices at individual instants do not change the integrals. Each rectangle has dimensionless unit area: (2 t0) times 1/(2 t0) = 1 and (10 t0) times 1/(10 t0) = 1. Let Mrecent be the stellar birth mass formed in the last 0 to 2 t0 and Molder the birth mass formed from 2 to 10 t0, in arbitrary positive mass unit M0. The summaries are A = Mrecent/(2 t0) and B = (Mrecent + Molder)/(10 t0), in M0/t0. Equal bin rates have masses (6, 24) M0 and averages (3, 3) M0/t0. The lower-recent case has masses (2, 28) M0 and averages (1, 3) M0/t0. The higher-recent case has masses (10, 20) M0 and averages (5, 3) M0/t0. All three cases have total birth mass 30 M0 over 10 t0, so B is 3 M0/t0 for all three, while A differs. No shape inside either age bin or instantaneous rate at the analysis epoch is specified. The kernel is an averaging weight, not a measured luminosity or birth-rate history. Real tracers have stellar-population, attenuation, sampling and calibration limits beyond this toy.

Use a tracer comparison as a physical test

Suppose a recent-sensitive indicator is weak relative to one with a longer response. Possible explanations include a falling recent birth rate, attenuation, escaped ionizing radiation, incomplete sampling or contamination of the other indicator. A useful analysis asks what additional observation distinguishes these possibilities.

Resolved age estimates, spectral line ratios, more bands and spatial comparisons can add constraints, but each must be interpreted through its own calibration and selection. More data can narrow alternatives without automatically establishing a unique answer.

Report the inferred formation quantity with its time weighting and mass convention. A rate averaged over an interval is different from total birth mass and from the current mass in stars and remnants. Changing among those quantities requires explicit accounting.

The next lesson follows that accounting through gas conversion, stellar return, inflow and outflow. It will keep a material budget separate from an energy budget and from causal claims about whether feedback triggered additional stars.

Worked examples

Compare histories II and III

Why does B agree while A differs by a factor of five?

Both histories contain total birth mass 30M0 over 10t0, so B = 3M0/t0.
Their recent masses are 2M0 and 10M0 across the same 2t0 interval.
A is therefore 1M0/t0 and 5M0/t0, reflecting different recent integrated birth mass.

The long summary retains only the full-window average; the shorter window distinguishes the recent portions.

Check the response normalization and units

What does a rectangle of height 1/(2t0) over a width 2t0 do to an averaged birth rate?

Its integral is [1/(2t0)](2t0) = 1.
Multiplying an age-dependent birth rate by that response and integrating in age preserves mass-per-time units.
For supplied recent birth mass 6M0, the result is 6M0/(2t0) = 3M0/t0.

It forms a finite-window average with rate units. It does not produce an instantaneous endpoint measurement.

Try the reasoning

Do equal recent and older bin-average rates prove an exactly constant rate at every time inside the bins?
Show the answer and reasoning
No.
The within-bin histories are unspecified; many functions have the same integrals.
Does disagreement between formation indicators uniquely measure the dust correction?
Show the answer and reasoning
No.
Time response, sampling, escape, attenuation and other emission sources can all contribute.
Is total stellar birth mass automatically the current retained stellar mass?
Show the answer and reasoning
No.
Stars can return material to the surroundings; the mass convention and evolutionary accounting matter.
Are the two supplied rectangles measured response functions for particular real observing bands?
Show the answer and reasoning
No.
They are invented normalized teaching summaries, with no real tracer timescale or calibration supplied.

What to carry forward

Formation indicators weight physical processes, stellar populations and ages differently.
Averaged birth rates require explicit time windows and a birth-mass convention.
A disagreement among tracers identifies a question to investigate, rather than one unique correction.

References

Teaching resources and research supporting this lesson’s explanations and examples.