John Pazmino
 NYSkies Astronomy Inc
 2009 June 6 initial
 2010 November 7 current
    On and off over the years the discussions at the NYSkies Seminars 
drifts to spectrometry. This arises usually from an article that bases 
its work on stellar spectra. In addition, dialog about cosmology 
revolves around spectra of galaxies. And spectra are the means of 
studying certain of the extrasolar planets. 
    Altho most home astronomers learn something about spectrometry 
thru textbooks and lectures, the subject remains one of the weaker 
parts of their expertise. I will not work over the textbook treatment 
but will cover selected features of spectrometry of particular 
importance for home astronomy. 
    I specificly leave out details of spectral classes, HR diagram, CM 
diagram, Doppler shift, atomic structure. These are well treated in 
the usual astronomy textbooks. I also skip a description of the 
spectrometer machine itself, which is hardly ever well explained in 
home astronomy litterature. However, the manufacturer's litterature 
has the detail you want or need.The emphasis here is on topics 
sometimes poorly explained elsewhere. 
Home spectrometry
    Home astronomers over the ages take part in their profession with 
equipment a decade or so behind that of their campus associates. It 
takes that long for a specialized device to be rebuilt in a commercial 
version within the abilities of the home astronomer. In other words, 
home astronomers in a given year can enjoy the same kind of devices 
once found only in the campus setting 10 and more years ago. 
    This does not mean that the new gadget is actually cheap or even 
affordable. It means that some one modified the design of the campus 
device so it can be manufactured in a smaller, lighter, simpler 
package within the skills and arts of the home astronomer. It could be 
quite expensive and only a few copies are sold each year. 
    One major exception is a spectrometer suitable for home operations. 
In spite of an enduring interest in the field there just isn't yet a 
competent, simple, mature spectrometer for the small telescope. The 
ones usually at sale are toys, ineptly designed and built, usable only 
for the Sun, good only for coarse images, unable to join with other 
devices, or grossly expensive. 
    Never the less it is possible to extract healthy pleasure from 
spectrometry, even tho you can not as yet take your own spectrograms. 
You avail of the spectra published on the websites of observatories 
and universities. You may collect them just as you collect pictures 
and videos and graphs and diagrams. 
    It is possible and relatively easy to capture low-dispersion 
spectra of bright stars. These are good for rough classification and 
to show that stars do emit varying amounts of light by wavelength. In 
a few cases the densest darkest absorption lines can be discerned. 
    The usual method is to place a transparent diffraction grating in 
front of the eyepiece, like a planetary filter, or the telescope, like 
a solar filter. 
    The grating produces spectra of every thing in the field of view, 
which can then be photographed in the conventional astrophoto manner. 
Such gratings are at sale for sensible prices and really do produce 
pleasing results. 
    However, except for coarse manipulation, the spectra are woefully 
below the quality for any serious examination. One possible exception 
is the spectrum of a bright nova where there may not be other better 
observations at that same instant. In this case almost any spectrum, 
to fill the record of the event, is valuable. 
    An other possibility is meteor spectrometry. If it is disgustingly 
tough to catch a meteor on a photograph, it is an order harder to 
secure a spectrogram of one. if you can catch a spectrum you may 
plausibly have the one and only analytical material for that meteor 
any where along its flight path. 
    Unlike for other sections of home astronomy there is a scanty 
litterature for spectrometry. The usual books are beyond most home 
astronomer's aptitude, being too technical or needing high-end physics 
and maths. 
    A couple extremely welcome pieces are worthy of study. The best 
one-stop book is 'Stars and their spectra' by Kaler from 1989. This 
work covers each spectral class with deep letterpress about each and 
has lavish illustrations. Being from the era when chemophotography was 
still used at observatories, many picture are of traditional spectra.  
    If you are into the technical aspects of life, try 'Astronomical 
spectroscopy' by Tennyson from 2005. The title has the traditional 
term and not the more general 'spectrometry'. This is a textbook with 
homework for each chapter. The very first chapter, the easiest of the 
lot, asks to work out the Doppler shift of a galaxy spectrum and 
compute the recession speed. Don't giggle. Some astronomers still 
explain the Hubble redshift as real motion of the source thru a static 
volume of space. 
    The newest book, as at spring 2009, is 'Stellar spectral 
classification' by Gray & Corbally. It's a blend of the first two, so 
you must be versed in astronomy and physics. Like Kaler, it steps thru 
the spectral classes with extensive letterpress. Its 500+ pages have a 
huge number of illustrations. Just about all spectra are densitometer 
tracings with few photographic spectra. The book has a website for 
digital spectra and a computer program to play with them. 
    The Contemporary Laboratory Exercises in Astronomy (CLEA) project 
has a lesson for spectral classification. Its computer program accepts 
spectrum data in a prescribed format and comes with 160ish standard 
stars to match unknown spectra against. CLEA's other lessons are 
wonderful, so check them out for your other astronomy interests. 
    You can cruise astronomy courses at universities, a huge number 
having webpages for their students. The quality varies all over the 
map from a simple list of classes and contacts to richly authored 
notes you can compile into a book. Squint at the course prerequisites 
to learn how much background you need to follow the lessons. 
    Precisa mente from the paucity of experience in spectrometry, 
websites by home astronomers tend to be awfully simple. They are 
mainly a rehash of astronomy textbooks and sometimes, due to honest 
defective comprehension, misleading.  
Some history 
    The term 'spectrum' seems to come from Newton who first made a 
study of the solar spectrum in the mid 1600s. He didn't know why such 
a colored display of light was created out of sunlight, but did notice 
that a candle flame yielded a different mix of colors in its spectrum. 
    Deliberate inquiry into spectra began in the early 1800s when 
better optical and mechanical skills were in hand. In 1802 Wollaston 
called attention to the dark lines cutting across the solar spectrum 
but had no clue to their nature or cause. He mused they were the 
divisions between the colors of the rainbow! 
    Fraunhofer in 1814 mapped many of the dark lines and gave them 
alphabetic designations. These are still in common use today as is the 
term 'Fraunhofer lines' for the lines generally. 
    Kirchoff uncovered three laws of spectroscopy in the mid 1850s, 
giving the terms 'continuum', 'dark-line', and 'bright-line'. These 
are elaborated below. Huggins in 1859 proved that certain nebulae were 
in fact made of a vapor. Until then it was supposed that the 
unresolved clouds will succumb to better and larger telescopes and be 
fragmented into constituent stars. 
    In 1863 Airy published the first systematic study of the Doppler 
shifts in star spectra. Rutherfurd perfected photography of spectra 
from his observatory in Rutherfurd Sq, Manhattan. Draper discovered 
oxygen in the solar spectrum from his home near Washington Sq, 
    Both Rutherfurd and Secchi separately in 1863 organized spectra 
into categories. Rutherfurd had three classes: Rigel-like, Sirius-
like, and Sun-like. Secchi had two, then four, based on prototypical 
stars. In 1870 he discovered carbon stars, now in a new class C. 
    By the 1870s it was obvious there was a similarity between the 
physics that governed spectra formation in laboratories and those in 
the heavens. This gave rise to the notion that the laws of nature are 
the same up there as here, 
    With that the science of astrophysics was born, which even today 
is often treated separately from plain astronomy. Many colleges today 
have a department of 'astronomy and astrophysics'. 
    In the 19th century a gadget to create a spectrum was solely a 
visual instrument. You looked into it and inspected the colored array 
of light. Hence, the device was called a spectroscope. Use of the 
instrument or the study of spectra was called spectroscopy. The two 
terms are still in wide use today even tho few astronomers actually 
look thru an eyepiece to observe the spectrum. 
    In the 1880s photography advanced to the capability to capture the 
spectroscope's image on film. In this combination the device is a 
spectrograph. With the spectrum frozen in the film, the spectrogram, 
it can be measured, shared, duplicated, preserved. 
    Modern spectrometry started in about 1890 at Harvard College 
Observatory under Pickering. His team of female astronomers -- 
Cannon, Fleming, Leavitt, Maury -- devised the spectral classes we 
still use today, altho in shuffled order. The work was financed 
initially by the Draper fund, leading to the Henry Draper catalog and 
extensions into the 20th century. This catalog is widely used today to 
select candidates for planetary stars. This is why so many are cited 
by their 'HD' numbers. 
    In the 20th century spectrometry is based on quantum physics, 
which describes the behavior of atoms and electrons under various 
environs of energy. 
    Spectrometry is the modern term for the study of spectra. The very 
device is now a spectrometer. There is essentially no longer any value 
to visually look at spectra thru an eyepiece, save for very bright 
targets during public or casual viewing. Photography on film is just 
about vanished from most astronomy, so we really don't generate 
spectrograms with a spectrograph. 
The spectrum
    For this article the spectrometer is a blackbox where a beam of 
light enters and a dispersed spectral beam leaves. The exit beam from 
a spectrometer is a linear or angular spread of wavelength. This 
dispersed beam is captured on some imaging medium, like chemophoto 
film or electronic sensing cell. On the image each wavelength is 
placed at a unique linear location where it can be isolated, examined, 
    The human eye & brain perceives the wavelengths as colors. 
Sunlight has a full range of colors, those of the rainbow. They are 
conventionally named red, orange, yellow, green, blue, indigo, and 
violet. I suspect the names were forced to number seven for good luck, 
since I haven't found anyone who honestly sees all seven colors in a 
solar spectrum or a rainbow. There really is no definite count of 
colors, not even in the rainbow. The colors blend smoothly into each 
other with no borders between them. 
    The spectrometer not only disperses the wavelengths but also 
widens the colored band laterally to make its details easier to 
inspect. This is done by the optics and mechanics in the unit. If 
there was no artificial broadening, the spectrum of a point or 
sngularly small target would be just a thin line. Its details would be 
very tough to discern. 
    The starlight enters thru a narrow slit to isolate as thin a beam 
as practical. Without the slit the image of the star at one wavelength 
will overlap that of an adjacent wavelength. The spectra from each 
unage would blend the texture, detail, and colors. 
    The spectrum consists of adjacent parallel images of the star 
wavelength by wavelength with as little overlap as practical as 
governed by the width of the entrance slit. The lateral breadth of the 
spectrum is the length of the slit as imaged by the spectrometer. In 
consequence, variations of light and dark within the spectrum show up 
as 'lines' orthogonal to the band of color. 
    In some spectrometers there is no optical image. The spectrum is 
recorded directly into a computer datafile to be later synthesized 
into a picture or, more usually, a graph of density vs wavelength. 
Spectrometers operating outside the visual range produce only the 
datafile. There aren't any lines as such. Never the less, regardless 
of the waveband covered by a given spectrum, distinct dark and light 
sections are still called 'lines'. 
    To capture spectra of nonstellar targets, essentially the same 
equipment, telescopes, apparatus, software are employed as for stellar 
spectrometry. The main difference is that for an angularly large 
target, each point along the slit over the target vreates its own 
spectrum. The result is a graduated spectrum from one end to the other 
of the slit, answering to each part of the target. 
Energy levels 
    I don't go thru the process of line formation here, that being 
better handled by regular physics and astronomy textbooks. In brief, 
for this piece, each atom has a set of definite specific energ states 
that its electrons may sit in. It's much like a set of shelfs here 
items can be placed on yy on a one or other shelf and have only 
distinct pottential energies relative to the ground. An item can not 
sit between shelfs. 
    The arrangement of energy states is part of quantum physics, first 
applied to atomic structure and spectrum formation by Bohr in 1913. 
The number of levels for an atom is infinite but the enrgy difference 
between levels narrows rapidly in the higher ones. After about the 7th 
or 8th level, the electron is so loosely held to the atom that it can 
just as well be free. The atom loses tht electron and becomes ionized. 
    Not only does each element have its peoper energy levels, but so 
does each isotope and each degree of ionization, where electrons are 
removed from the atom. 
   An electron may abosrb outside energy, applied by heat or 
collision, for example, if and only if the amount is exactly that to 
shift the electron to a highrer energy level. Any other amount is 
passed thru and the atom is reansparent to it.
    The electron doesn't like being in a high energy state and will 
drop to a lower one, In doing so it emits a photon of the exact energ 
difference of the two levels. 
    Altho an atom has an infinite number of energy levels, they are 
NOT all occupied by electrons at once. An oxygen atom has 7 electrons 
and fils only 7 of its energy levels. The others are empty but can be 
occupied by electrons shifted to them. 
    For an atom to create spectral lines, it must actually have at 
least one electron with it. It is the exchange of energy with an 
electron as it jumps from one energy level within the atom to an other 
that creates the line. 
    If the atom is thoroly ionized to have no more electrons, it no 
longer produces spectral lines. Very hot stars have no lines from 
hydrogen, a puzzle to early astronomers who were sure hydrogen was in 
the stars. The hydrogen was heated to ionization, losing its one 
electron. There was no electron to jump among the hydrogen energy 
levels to produce spectral lines. 
   The higher the temperature, the higher energy level within the atom 
an electron may be excited to. Transitions between that level and 
those of lower energy can produce lines. By noting which lines are 
present in the spectrum, the highest energy level among them can be 
figured out. 
    In turn this highest energy level fixes the highest temperature a 
star may have. With a given atom having only discrete energy levels, 
the lines from many different atoms are examined to box in the star's 
    Energy levels are commonly called 'orbits' after the original Bohr 
study of the hydrogen atom. This is a bit misleading, harking back to 
the solar system model of atoms of most high scholl science classes. 
First, the atom is a 3D structure, so at least the better term is 
'shell'. In fact the electrron deployment around the nncleus is 
compplex and the energy levels are named in a letter-number scheme. 
    Light waves have an incredible short length on the human scale. 
It's about 500 nanometers or 1/2 micron. An older unit, still in wide 
use, is the Angstrom, 1/10 nanometer. Light wavelengths are at about 
5000 Angstroms. Context resolves ambiguity but it is best to always 
explicitly specify the unit of measure. 
    1 Angstrom = 1/10 nanometer
    1 nanometer = 10 Angstrom 
    Astronomers in the visual zone of the spectrum don't cite 
frequency, like in the radio bands. Nor do they state energy, like for 
gamma-rays and X-rays. This is merely an accident of history. 
    Frequency, energy, and wavelength are related thru the formulae
  (lightspeed) = (wavelength) * (frequency) 
  (energy) = {Planck constant) * (lightspeed) / (wavelength) 
  (energy) = {Planck constant) * (frequency) 
The Planck constant is the 'big' one, with 2*pi already included, 
equal to 6.626e-34J.s. If the 'small' value is to hand, multiply it by 
(2*pi). Lightspeed is 2.998e+8m/s. 
    When the wavelength is in Angstroms, some astronomers leave out 
the unit, like '... the line at 4228 is ...'. Other astronomers use 
'lambda' for Angstrom rather than the symbol 'A' with a 'o' hat. In 
ASCII text, plain 'A' is acceptable. 
Blackbody radiation 
    One of the royal cockups in most astronomy textbooks is to treat 
LUMINOUS energy, light, as if it was the TOTAL radiation emitted by a 
star. The author applies to LIGHT the laws of Planck radiation. These 
laws are valid ONLY for the ENTIRE wavelength range from zero to 
infinity and NOT to a section within that range, like the visual band. 
    This weird situation was a necessity in the years before we could 
explore radiation outside the visual range. Our atmosphere and early 
instruments did not allow such observations. It took the space age to 
enable us to bring spectrometers out of the atmosphere and receive the 
full wavelength range. 
    We knew stars were Planck emissors, so we could estimate how much 
of the total radiation output we were capturing in the optical band. 
It was a major portion because of how Planck radiators work in the 
temperature of thousands to tens of thousands of degrees, that of 
    In fact, we do add a puffup factor to estimate the total radiation 
output based on the luminous portion er see. This is the bolometric 
correction, expressed as a magnitude increment. It NEVER was meant to 
show how bright the star would look if we could somehow see in all 
wavelengths. There really is no such a concept in nature. It's like 
somehow converting music into a visible image to show what it looks 
like if our eyes could perceive sound. 
    For most stars the bolometric correction is a fractional magnitude, 
meaning that only a minor amount of radiation falls beyond the violet 
and red ends of the spectrum. For very hot or very cold stars the 
correction can be several magnitudes. This means that most radiation 
is beyond the visual spectrum and must be accounted for before using 
the radiation laws. 
    As it turns out, for MOST stars the BULK of its Planck emission IS 
in the visual band, up to maybe 95 percent. Hence, with exceptions we 
must be mindful of, we can apply Planck rules on just the visible 
radiation with the caveat that the results can be only approximate. 
    Occasionally a proposal comes along to redefine spectral classes 
from the total radiation profile, not just the visual part. None was 
widely accepted. I can see in the future that some thing has to be 
done to recognize the nonvisual energy, in as much as we now routinely 
observe that energy from space-based observatories. 
    A severe complication, not anticipated before the space age, is 
that outside the visual range, most of the radiation is NOT blackbody 
radiation. It is some other nonthermal radiation against which the 
Planck laws are utterly invalid. 
    Sources that generate radiation by temperature, a blackbody or 
Planck emitter, sends out radiation of all wavelength. Those within 
the limits of human perception are light. A spectrum taken of a 
blackbody source is a continuous band of color shading from red thru 
violet. The continuous background in a typical stellar spectrum is 
generated by the blackbody radiation from the star's photosphere. This 
is the continuum of the spectrum. 
    The continuum has a wavelength of maximum emission somewhere in 
the spectrum. For most stars this peak wavelength is within the visual 
spectrum. Extremely hot or cold stars peak beyond the visual band. 
    If the peak wavelength of the spectrum is determined, the 
temperature comes directly from it by applying Wien's law, from Planck 
radiation physics. The peak wavelength is inversely proportional to 
    Te blend of emission across the continuum excites the ryr & brain 
to perceive a color. Since color perception varies so widely in 
humnans, there is NO precribed color for each temperature. The colors 
actually seen in starlight are only loosely related to temperature and 
can be, erm, colored by proximity to other stars. Double stars show 
colors enhanced by constrast between the component stars. 
System of units 
    Stricta mente temperature in science is in degrees Kelvin, a must 
for working with the radiation formulae. The offset of centigrade, or 
Celsius, from Kelvin is only 273 degree. For all but the very coldest 
of stars, the percent error is negligible. That's why star temperaturs 
can be cited as eeither Celsius or kelvin interchangeably. Recall that 
        (temp in C) = (temp in K) - (273) 
        (temp in K) = (temp in C) + (273) 
    Many authors don't state the unit, C or K. You must figure out 
what is intended from the text or math. You may have to do a sanity 
test by plugging the given numbers into one of the equations. For 
purely descriptive functions, it doesn't matter. 
    As for other measurements, metrics is the system of the realm. 
Oldstyle is a relic now found only in legacy litterature. There is 
still prevalent employment of the CGS system, the antecedent of the 
MKS or SI units. Sometimes the units are mixed in the same work. Such 
loose handling of measures can be a irritation at first, but you soon 
learn to slide the decimal point to rectify the units to a single 
    You will encounter other units peculiar to atomic and quantum 
physics. Some are older metric units while others are convenient 
special ones. It helps to have an atomic physics textbook handy to 
decipher these extra units. 
Dark-line spectrum 
    When a blackbody source shines thru a vapor or gas its spectrum 
has dark lines across it. These are produced by the vapor's absorption 
of specific wavelengths from the source radiation. Each chemical 
element and siotope in the vapor generates its own unique set of dark 
lines. It didn't take long to use this fact in chemical analysis of 
industrial materials and criminology. This was done by the mid 1800s, 
even tho we then had no inkling why such spectra were so unique. 
    For the most part, stars give off a dark-line spectrum, also 
called an absorption spectrum. The photosphere emits Planck radiation 
to form the continuum. The transparent atmosphere absorbs specific 
wavelengths to form the dark lines. 
    Only a couple percent of the total blackbody radiation is 
absorbed. As a first approximation, a star is a pure blackbody emissor 
and the Planck laws apply to it with little error. 
    Cold stars can have many dense dark lines that obscure a 
significant portion of the continuum. This spoils a pure blackbody 
treatment. In fact, in some cases the continuum is so smothered that 
the spectraum looks as if there really is none. 
    The lines are not completely devoid of radiation. They are really 
a dark gray, with some radiation in them. Once the radiation is 
absorbd, it is reemitted as photons of the same wavelength. The 
emitted photons may leave the atom in any direction, not only forward 
with the blackbody radiation. There is a net deficit of light at the 
    The smmall amount reemitted light sent forward can be captured for 
special studies within its wavelength. An H-alpha solar filter does 
this by examining the Sun within the hydrogen-alpha line due to 
    Because this is a very dark line, with little residual radiation, 
the image thru the filter is dark compared to the white-light 
[fitered] view of the Sun. It is commonly viewed under a hood to blcok 
ambient light from the eyes. 
Bright-line spectrum 
     When a gas or vapor is itself made to shine, like by an electric 
current, the generated spectrum consists of specific wavelengths with 
no continuum. The abosrbed energy from the electricity is not luminous 
so the reemitted photons are seen clearly. These wavelengths show up 
as bright lines, unique to each element in the vapor. Such a spectrum 
is a bright-line or emission spectrum and is exploited for chemical 
analysis, just like the dark-line spectrum. 
    For a given composition of gas the dark-line and bright-line 
spectra have the lines at the same wavelengths. Extensive experiments 
in the 19th century showed that the only factor in the occurrence,  
placement, density of the lines is the chemical makeup of the gas. 
    There are other factors but they were too weak to discern under 
19th century conditions. They include pressure, motion and agitation 
in the gas, and magnetic fields. Altho by the turn of the 20th century 
these other influences on a spectrum were studied in a laboratory, 
they were not confirmed in star spectra until the early 20th century. 
    Few stars generate a bright-line spectrum. A nebula does, being 
that it is a cloud of gas energized by nearby stars to shine. 
    A bright-line spectrum of an angularly tiny nebula can be studied 
without a slit. The target forms separate images of itself at the 
wavelengths of its bright lines. The structure of these images varies 
because the element associated with each line is present in different 
parts of the target. 
    This technique is done to pick out a tiny globular nebula in a 
field of stars. Place a transpatrent diffraction grating, 'rainbow 
plastic', over the eyepiece. Stars give off continuum spectra. Their 
dakr lines are too hard to see. The bebula gives off a series of dots, 
its bright lines. 
    The energy levels in an atom are a function of the number of both 
the neutrons and electrons in the atom. Varying the neutrons alters 
the energy levels a smidgeon so it is possible in high dispersion 
spectra to distinguish the isotopes. 
    A greater modulation of the energy levels is caused when electrons 
are removed from the atom by sufficient energy, An atom with electrons 
missing is an ion and is ionized. The remaining electrons realign into 
new levels. The lines they now produce are different, making it easier 
to uncover ionized states among the atoms. 
    The highest ionization state of an atom in a star, the ones with 
the most electrons torn off, is an index of the star's temperature. 
This is the way we get the temperature when there is little 
conyyinuum, like when the spectrum is filled with too many dark lines. 
    Ions are named in two ways, one from chemistry and one from 
physics. Both are in circulation among astronomers, reflecting their 
background and upbringing. In the chemical notation the number of 
excess positive charges is stated, resulting from removing the same 
number of negatively-chaarged electrons. 
    An oxygen atom with two electrons removed is O++. There are now 
two extra plus charges, the protons, no longer paired with electrons. 
For high ionization states, the explicit number is given to avoid 
counting lots of '+'s. Fe+7  is an iron atom with 7 electrons removed. 
    In the physics notation the energy state is given as a Roman 
number. The neutral atom with all electrons is in state #1. He-I is 
neutral helium with both of its electrons. When one electron is pulled 
off, the atom is in its 2nd energy state. Mg-II is magnesium with one 
electron lost. 
    Roman numbers are compact enough to always give the actual number 
of the energy state like Co-XI for cobalt in its eleventh state with 
ten lost electrons. The physics notation is ONE GREATER than the 
number of lost electrons. 
    Here are examples of the two notations for element 'pazminium'. 
        chemical | physical | electrons 
        Pz       | Pz-I     | none, neutral atom  
        Pz++     | Pz-III   | two removed 
        Pz+++++  | Pz-VI    | five removed 
        Pz+12    | Pz-XIII  | twelve removed 
    In general the ionization states present in a spectrum depends on 
the star's temperature. In the colder stars, we see molecules and 
neutral atoms. In higher temperatures we see atoms in their first 
ionization for electrons weakly held to the atom. Higher still we see 
the stronger-bonded electrons removed and multi-level ionizations. At 
the highest temperatures the atoms are stripped of all or most of 
their electrons. 
Line designation 
    Spectrum lines are named from legacy or by merely citing their 
wavelength. Either Angstroms or nanometers may be used, being careful 
to state which. A nanometer value mistaken for Angstrom puts the line 
in the far ultraviolet. An Angstrom value taken as nanometers places 
the line in mid infrared. 
    Lines can also be named for the element and ionization state that 
produced it, like O-II (or O+) for a line of singly ionized oxygen. 
This is ambiguous because there are many lines due to a given 
ionization for an atom. Context resolves the ambiquity. 
    An other way is to gather lines of a given element into a series. 
The Balmer series for hydrogen lines has Greek letters like the 
hydrogen-alpha line or hydrogen-gamma line. 
    A specialized naming is the actual energy levels associated with 
the line. The notation is taken from quantum physics as a mix of 
letters and numbers. This labeling is mainly for detailed work with 
the lines of a specific element, each of which is produced by a 
electron transition between different energy levels. 
    An older method, still in common use for solar spectra, is the 
Fraunhofer Latin letters. These run from red to blue because the very 
first spectrum diagrams placed the red end at the left. The A and B 
lines are created in the Earth's air by molecular oxygen. The H and K 
lines come from calcium and the D line is from sodium. 
    Only the major lines are so named, leaving myriads more to be 
designated by the other methods. Fraunhofer letters are almost 
exclusively reserved for spectra of the Sun, being rarely, if ever, 
cited for other targets. 
    As far as I know lines are no longer named for persons. We still 
speak of the Balmer, Lyman, anf other series, but these were named 
decades ago. Within each series the lines are given Greek letters like 
    In a given spectrum example the lines may be labeled by a mix of 
methods. There is no attempt to clean up the designations. 
Spurious lines 
    Not all the structure in a star's spectrum originate in the star. 
The starlight can be affected by any thing along the path from the 
star to us. A gas cloud between the star and us adds its own lines 
into the spectrum. These are the interstellar lines. They are 
important for probing material that would be invisible except for the 
luck of a star shining thru it. 
     An interstellar line is revealed by having atoms in energy levels 
much too low to survive at the star's temperature and by showing a 
motion out of step with the star's own motion. 
    When the starlight passes thru Earth's air, more lines are 
generated. These lines are recognized for being of atmospheric 
molecules and varying with the star's altitude above the horizon. 
These are the telluric (TEH-luh-rikk) lines, from 'tellus, telluris', 
'mother earth'. 
    In decades past the telluric lines were mostly absorptions, making 
dark lines in the spectrum. Fraunhofer's A and B lines are caused by 
molecular oxygen in our air, not from anything going on in the Sun.
    In recent decades telluric lines are showing up as bright lines. 
Light is emitted from Earth in luminous graffiti. As lamps shift away 
from incandescent bulbs to energy-efficient alternatives, the luminous 
graffiti tends to be a bright line emission. The newer lamps convert a 
greater fraction of input electric into light at a select few 
    With subdued luminous graffiti the bright lines are helpful as a 
'ruler' of known wavelengths to measure the positions of the star's 
own lines. The more offensive graffiti makes dense lines that obscure 
the star's features. 
    There is an other kind of line, not spurious as such but at first 
it looks suspicious. This is the 'forbidden' line, from an earlier era 
of spectrometry. These are lines formed by physical conditions that 
once were not feasible in a laboratory, They were 'forbidden'.
    By the late 20th century, labs advanced so that any astronomicly 
important environment, even that near the bigbang, can be reproduced 
on Earth. Hence, there are no longer any forbidden lines. 
Rendering spectra 
    Historicly spectra were photographed by uniting a camera with a 
spectroscope. In time the two were built into one unit as the 
spectrograph. The image is surprisingly small, judging by the large 
reproductions in textbooks. A spectrum may be only a few millimeters 
long and is examined with a microscope. Measurements are made on a 
traveling stage with micrometer controls. 
    The image is a B&W picture, there being no need to capture a 
colored picture. Color film for spectrometry was limited to taking 
publicity or news pictures. Color film has many layers of emulsion 
that can diffuse delicate texture in the spectrum. B&W has a single 
layer onto which all the spectral details are recorded. 
    When densitometers were improved, they were used to scan the 
photograph and produce a tracing or profile of density, opacity, 
darkness against wavelength. This gave a more detailed and textured 
spectrum to work with. it is far easier to study and compare spectra 
with densitometer plots than with the raw photographs. 
    The densitometer recorded wavelength by wavelength the flux of a 
pinpoint of light as it traversed the spectrogram. One stupendous 
advantage of the plot is that it was quick and easy to photocopy and 
circulate without the hassle of chemophoto darkroom work. 
    It didn't take long to convert the densitometer readings into 
digital form. Each point in the spectrum became a record in the file 
with two fields, a wavelength and a density value. 
    Once in digital form in a datafile it was trivial to write 
computer programs to read the data and do various analyses on them. 
The datafile can be transmitted by email or placed on a website so 
other astronomers can study it. 
    Today just about all spectra are generated as numerical data. If 
you want a 'photograph', many spectrum programs will synthesize a 
shaded rendering of the densitometer tracing. This is pura mente a 
schematic picture, not valid in itself for analysis. 
Normalized spectra
    A raw densitometer plot of a star includes the continuum as an 
envelope over the graph. This is a good way to illustrate hoe 
temprature coorelats to the spectral lines. Hot stars have continua 
peaking in the blue/violet or beyond. Middling stars peak within the 
optical band. The Sun's continuum peaks close to the peak of human eye 
spensitivity, one major fact considered in anthropolgy sutdies. 
    Because it is impossible to calibrate all spectra to a uniform 
standard density, Spectra are commonly normalized for themself. The 
continuum is assigned unity density. All other points have density 
relative to one. In this rendition, the continuum is a horizontal line 
across the plot. 
    A profile of a dark-line spectrum has dips, gorges, valleys for 
the lines. They are denser, darker, than the continuum. A bright-line 
spectrum has spikes, peaks, mounds above the continuum, being brighter 
than it. 
    For cold stars whose continuum is obscured by crowded lines, the 
densitometer tracing is left in raw form. The tracings have a steep 
upward slope from violet to red and line densities can not so simply 
be compared against each other. 
Dispersion and Resolution 
    The ability of a spectrometer to separate wavelengths is is 
dispersion. The value s wither in fractional angstroms or in its 
reciprocal. A spectrometer that discriminates 1/2,000th of an Anstrom 
has dispersion of either 1/2000, 0,.0005, or, as reciprocal, 2000. 
    This is sometimes made out to be equivalent to the resolution of 
the spectrometer. This is the Angtrom/millimeter imaged on the film or 
sensor cell. Altho the two are correlated, the dispersion is the 
ultimate amount of detail in the spectrum while resolution is the 
ability of the image to capture that detail. 
    The resolution of a spectrum is the Angstrom per millimeter on the 
image. Because sensors when matched to the spectrometer have pixrl 
size less than the dispersion rating, the rendered image captures all 
of the detail and texture from the spectrum. 
    Magnifying this image does not increase the detail any more than 
magnifying an ordinary photograph adds more detail. The texture in the 
image is merely made bigger and likely more blurred and diffuse. 
    Home astronomers are happy to get a spectrogram, most still 
preferring a photographic rendition, with resolution of tens of A/mm. 
At this level there is an amazing amount of texture, enough to excite 
and satisfy the home astronomer's interest. 
    A resolution of a few A/mm is sufficient for classifying spectra, 
like in an astronomy course. For important work resolution of 
hundredths of A/mm is required. 
Spectral classes 
    The classification of spectra began in the mid 19th century and 
was well established by the end of that century. The method is based 
on the Harvard College system where each category of spectrum is 
assigned a Latin letter. The original sequence was alphabetic in 
descending strength of the hydrogen lines in the spectrum. This scheme 
ran from A, densest hydrogen lines, to O, absence of thee lines. 
    In 1900 the concept of temperature as a prime parameter of stars 
was used to reorganize the spectral classes. Class O was the hottest 
of the stars and M was the coldest. Some classes were dropped or 
combined as mistakes or minor variations of others. What's more, when 
the sequence of classes was shuffled the letters were fixed. That's 
how we end up today with the jumbled set O, B, A, F, G, K, M. 
    In the 1960s and later a few new classes were added, some taking 
letters that were dropped decades ago: C, L, S, T, W. Some astronomer 
balk at making up more classes and keep just the historical set of O 
thru M. 
    Temperature produces the lines according to the energy states  of 
the atoms. The hottest stars had lines from ionized atoms. Moderate 
stars had lines from neutral atoms. The colder stars included lines 
from molecules. A first-cut segregation of spectra is by seeing which 
energy states of atoms have lines in the spectrum. 
    The range of temperature, in Kelvin, for the classes is set out 
here. The colors are schematic, alluding to the mixture of wavelengths 
that the eye blends into a composite hue. Also given is the range of 
color index, explained below, for each spectral class. 
          | temperature   | color index   | color 
        O | 30,000-45,000 | -0.57 - -0.49 | blue 
        B | 10,600-29,000 | -0.48 - -0.04 | blue-white 
        A |  7.400-10,500 | -0.03 - +0.28 | white 
        F |  6,000- 7,300 | -0.27 - +0.49 | yellow-white 
        G |  5,400- 5,900 | +0.50 - +0.63 | yellow 
        K |  4,100- 5,300 | +0.64 - +1.07 | orange 
        M |  2,500- 4,000 | +1.08 - +2.19 | red 
    There are stars hotter than 45,000K and colder than 2,500K but 
they fall outside the usual scheme of spectral classes. Attempts to 
project the classification system into the nonvisual range haven't 
been very favorably accepted. 
    These classes are coarse and do not fully describe the many 
varieties of star. In time each was divided into subclasses 0 thru 9. 
Textbooks may say that each class has 10 divisions. In fact, some 
classes have less and some more than 10. The latter go to decimal 
Color index
    An early, and still useful, way to obtain a first-cut spectral 
classification is to image the star in two distinct wavebands. A 
comparison of the apparent magnitude of the star in the two images is 
correlates with the star's blackbody curve. In essence, this is a two-
color colorimetry method, suitable for very faint targets or large 
numbers of them. In general, a target will impress different 
illuminations in the two bands because the bands cut the blackbody 
curve at two different illuminations. 
    By history the comparison is made in the B (blue) and V (visual, 
in the yellow) bands of the astronomy photometry system. The color 
index itself is the subtraction of the V magnitude from the B 
magnitude, B - V. Hence, the index is widely called the 'B-V index'. 
    The subtraction is algebraic, minding that the magnitude value 
decreases with increase of the star's illumination, the number of 
photons impressing the image. 
    The color index method works because a blackbody curve is uniquely 
fixed by two points on it. Only one curve can satisfy the given points 
and there is only one temperature, spectral class, that matches the 
points. In fact, some HR diagrams are plotted with color index and not 
spectral class or temperature. 
    The bands are zeroed so that a star of spectral class A0 has a 
color index of 0.0. That is, the illumination received thru the blue 
and visual bands are the same, or the spectroillumination at these 
wavelengths are the same on the star's blackbody curve. 
    Recall that the peak of blackbody radiation shifts to the blue 
side of the spectrum, short wavelengths, with increase of temperature. 
This causes an imbalance between the spectroilluminations at the B and 
V wavelengths and their difference is no longer zero. 
    A star of higher  temperature than A0 will have less energy in the 
V band than in the B band so its color index has a negative value. 
Stars cooler than A0 have more energy in the V than B, making their 
color indices positive. 
    Please mind well that color index is NOT the bolometric 
correction. Both are stated as a magnitude and are based on the Planck 
curve of the star. However, they serve entirely separate functions. 
    In the table above of temperature and color, the range of B-V 
index is also given for each spectral class. 
Star colors 
    It is common in texts to assign colors to the spectral classes, 
like an F star is yellow-white in overall color. The color perceived 
in a star is so tangled in extraneous factors that it's a miracle that 
there is any time and energy spent on describing stellar colorations. 
In the 19th century, before blackbody radiation was fully explained, 
there was a all-points program to carefully assess star colors and 
their changes with time. Color patches and palettes were published to 
compare with the stars and record the matching color. 
    Eyesight, temperament, weather, air pollution, ambient light and 
noise, fatigue, health, diet, proximity of other colored stars (double 
stars) all play substantial parts in generating a perceived color. It 
is rather hopeless to insist that you must see Arcturus as an orange 
star because it has a K class spectrum. 
    If you do not see Arcturus as orange, you're OK. There's no way 
you can be taught to see Arcturus as orange if your physiological 
makeup and ambient circumstances don't let you. Yet in home astronomy 
litterature you read, '... a little higher up is a blue star ...'. It 
could look blue to you, or look blue-white, purple, blue-green, or 
plain white. 
    The one color that does seem definite in stars is red. Most people 
can pick out the red stars in a star field. Some stars are so red they 
resemble a ruby, garnet, other red gemstone, or blood. Some 19th and 
early 20th century maps marked red stars with 'r' at their symbols. 
    The colors assigned to the spectral classes are purely schematic. 
They are used for coloring stars on charts computers or tinting their 
images in planetaria. 
Morgan-Keenan system
    The spectral classes of today is based on the work of Morgan and 
Keenan in 1943, with evolution since then. This scheme is an upgrade 
of the system developed at Harvard College. Hence, overall, if you 
read litterature from the entire 20th century thru today the notation 
for the spectral types is pretty much the same. A star called 'B5' in 
1920 is likely still classed as B5 star today, barring mutations of 
the star or errors in classifying it. 
    By the way, stars can alter their radiation regime on timescale of 
days to decades. Within your lifetime you can really see evolution in 
certain stars in realtime. 
    This Morgan-Keenan, or MK, system prescribes definitions of each 
class based on the structure of the spectrum. It keeps the traditional 
letters and numbered subtypes, but gives them firmer and more positive 
criteria. Factors like the ratio of strength among the lines, the 
presence and absence of certain lines, the width or breadth of the 
lines are used. 
    The classification is purely based on the spectrum features, not 
theory or cause. If a spectral line is later found to have a new 
explanation, that is laid into the description of the class. The star 
is not reviewed for a possible new class. 
    To calibrate and enforce the MK system a set of standard stars was 
assessed. These were selected from all over the heavens. Each was 
assigned the spectral class according to the MK criteria. The target 
star is compared against the standard stars to yield its spectral 
    You don't have to simultaneously observe the standard stars. Their 
spectra are documented in both print and digital form. Computer 
programs can align the standard spectra with that of the target to 
facilitate the comparison. 
    The original work in the 1940s was done in the blue end of the 
spectrum because at that time photographic films were only weakly 
sensitive to the yellow and red portions. 
    After World War II Kodak's newly designed spectroscopic film and 
today's digital imaging capture the entire spectrum equally well. 
Never the less, spectral classification is still performed almost 
entirely on just the blue side. 
    By good luck the blue end has enough astronomicly important lines 
that there was no real push to expand the class criteria to the rest 
of the spectrum. 
    The irregular spacing by temperture of the spectral divisions can 
make an HR diagram look distorted. The usual scheme is to mark off 
spectral classes at equal intervals on the X-axis, each divided into 
ten parts. 
    This ignores the varying number of defined steps in each class. 
SOme classes have fewer or mowre than ten steps. As stars are plotted, 
certain subclasses are missed, leaving gaps in the graph. The curve 
thru the other points may be crooked. 
    A better plot has temperature or color index on the X-axis. The 
curves on the graph are smoother and more continuous. 
Early and late
    The terms 'early' and 'late' are commonly used in spectrometry. 
These come from the era when stars were thought to be born at a high 
temperature and then gradually cool off over their lifespan. 
    We know know that a star changes temperature radicly over its 
lifespan in a complicated way, not by a simple hot-to-cold path. Yet 
the terms early and late endure. 
    Today they are about synonymous with temperature trend. Early 
stars are hotter stars; late, colder. A hot star, spectral classes O, 
B, and A, are 'early' stars. Those in classes G, K, M are 'late'. The 
terms also mean relative location within a class, as 'early K' means a 
stars nearer to G, hotter, than to M, colder. 
Luminosity class
    By the 1920s it was realized that a star can have the same 
spectral class but very different luminosity. More over, the more 
luminous stars tend also to be the larger ones in diameter or volume. 
    A second dimension of classification was developed to show not 
only the spectral class, based on temperature, but also the 
luminosity, say in solar units. 
    The luminosity is revealed by a detailed inspection of the 
spectrum. The criteria for the luminosity classes are set within each 
spectral class. There is no global rule that applies directly without 
first knowing the spectral class. 
    The classes are Roman numerals as follows, with the addition of a 
        0   - hypergiant stars, most luminous 
        I   - supergiant stars 
        II  - bright giant stars 
        III - giant stars 
        IV  - subgiants, just above Main Sequence 
        V   - Main Sequence and dwarfs 
        VI  - subdwarfs, below Main Sequence (little used) 
    The luminosity is appended to the spectral class as in G-II for a 
giant G star. 
    Luminosity classes are best differentiated within class F thru M. 
earlier (hotter) than F the luminosity classes are so compressed that 
many astronomers skip them. 
Extravisual band 
    The visual part of the spectrum spans about 380nm-760nm (3800A- 
7600A). The exact value depends on the spectrometer and the absorbing 
quality of the air above it. One way around the latter is to place the 
spectrometer at a high elevation, where the thinner and drier air 
blocks less radiation. Spectrometers are routinely part of payloads on 
balloons and satellites to get higher above or outside the atmosphere. 
    From time to time there are proposals to define classes outside 
the visual band, at least into the infrared and ultraviolet zone. None 
are generally recognized today, but are sometimes employed. One 
problem is that the concept of a 'spectrum' rests fundamentally on the 
eye's reception of light from the source. 
    An other factor is that in the infrared and ultraviolet bands, 
spectra can include major fractions of nonthermal features. The lines 
are increasing produced by other than electron energy transitions. In 
the infrared, the lines are mostly made by molecules, whose physics 
and chemistry are quite different from pure elements. 
    There seems to be no obvious or acceptable way to define criteria 
as a logical and consistent extrapolation of the MK scheme beyond the 
visual zone. 
    A crude form of spectrometry is performed thru colorimetry. This 
is the capture of regular photos or digital images of stars thru color 
separation filters. A picture is taken thru each filter and the density 
of the star image is recorded. 
    The several densities values correspond to points on the spectrum's 
continuum. This in turn fixes the shape of the continuum and thus the 
temperature of the star. This finally correlates to spectral class. 
This method is crude but it works when there is no spectrometer, the 
target is too faint for the one to hand, or there are too many targets 
to  capture individual spectra. 
    The filters have to be those for astronomical photometry, not for 
lithography, graphic arts, photography. These are called RGB filters 
after the colors they show when held up to daylight. You can get the 
latter for very cheap in a graphic arts shop but their wavebands do 
not at all correspond to those for astronomy purposes. 
    The correct set can cost a bit. Follow the ads for astronomy 
imaging supplies and accessories. The set, at least for home astronomy, 
has filters for the infrared, red, yellow or visual, blue, violet, and 
ultraviolet. They have central peak transmissions in the middle of 
thee colors and width of bandpass that overlap the adjacent filters. 
    Some skill and after-work is required to get full intelligence out 
of the colorimetric images, but they are within the ability of home 
astronomers. In fact, colorimetry may be the best way to capture 
information about a sudden or rapid event like a nova or gamma-ray 
    It may be that a special set of colorimetric filters was designed 
to match the spectroresponse of your peculiar imaging device. If so 
you better use only that set. Also be mindful to avoid using the 
filters with some other device.
    Spectrometry is far more within reach of the home astronomer today 
than it ever was in the past. Yes, it is still not quite feasible to 
take your own spectrograms with your own instrument. On the other 
hand, the rapid distribution of digital spectra, specially for targets 
in the current news, and the free programs for playing the spectral 
files make spectrometry an enjoyable and educational pursuit. 
    On the experimental side of astronomy, spectrometry was in the 
past a tough subject to learn because we had only the couple sample 
pictures in texts and we had to accept the author's word about what 
they showed. By manipulating digital spectra on your computer you far 
better can appreciate the nature and value of stellar spectra. 
    On the other hand, it is still true that you better bone up on 
atomic physics, thermodynamics, gasodynamics, relativity, optics, 
atmospherics. These and other fields remain necessary for a full 
appreciation of spectrometry.