by Martin Kölling, 2015. Version: May 6, 2025  found on:  https://www.sedgeochem.uni-bremen.de/kiloyears.html

If we want to express thousand meters, we call it "km" - a lowercase "k" for the multiplier 1000 and "m" as the generally accepted SI unit of length. If we want to express that two points are 1000 m apart, we would say one is 1 km from the other. If something is 25000 m away, we would say its 25 km away. And if a highway exit is at highway km 34 we would say its at highway km 34, no matter if it is only 1 km away. Very simple, very clear. If something is 20 km to the left, we might use -20 km and +20 km would be to the right e.g. like in a normal cartesian coordinate system. If a plane flies at 8840m, we could also say it flies 8.84 km high, and everybody would understand (ok pilots would call it 29000 ft, but this is a different story). If a borehole is 2000 m deep, you could call it 2 km deep. Nobody would dare to come up with a unit that is "kma" for km apart or "kmd" for km deep. And I have not seen people use "me", "mr" or "met" instead of "m" for meter. Its just "m". Clearly not a capital "M" which stands for Mega or the multiplier 10⁶. The only complication is, that "m" also stands for the multiplier 10^-3, but everybody seems to be perfectly fine with "mm" meaning millimeters or 0.001m - not m².  But then we just found the first evidence of using two multipliers in one unit:  Mkm² (Müller & Joos,2020) which stands for million square kilometers. The authors mean 10⁶ km² or 10¹² m² or  Mm² (which is unusual I have to say - but in line with SI). But if Mkm² is allowed, we might as well use kk(km²) and instead of TB for TeraByte we could use MMB or M²B.  Hmmm (that is just representing me thinking - not Henry times micro meters) !

Yet, when it comes to express thousand years scientifically, it gets complicated (maybe we scientists get complicated): It seems like a competition to find the one non-standard unit that nobody ever used before. We have actually found more than seventy (!) ways (and counting) of expressing thousand years (ago) in a scientific paper. And most of these different versions have been used in major peer reviewed journals.

There also seems to be some confusion about capitalizing units. Even when the title of axis should be capitalized as many journals request, this does not mean that units should swap to upper case !  "k" stands for kilo but "K" stands for Kelvin!
So a simple correct age axis label is "Age [cal ka BP]"  - not " Cal Kyr B.P.". And there is no plural form of a unit. Nobody would think of using 3 ms instead of 3 m for 3 meters - ms stands for millisecond. So its 300 kyr or 300 ka, but not 300 kyrs or 300 kai for kiloyears or kiloanni.

Here is a list of what scientists write (or what reviewers and editors force them to write) when they mean thousand years. The rating is my personal opinion:

Just when we thought, every single possible variety had been used, Science came up with a new spelling in 2015: "KYA" which most likely meant "kiloyears ago". (Strictly spoken, "KYA" means "Kelvin Yotta Ampere", with Yotta standing for the multiplier 10²⁴). Nobody knows how this all uppercase spelling of a unit that is unusual already in its lowercase version (kya) could have been agreed on by dozens of co-authors and how it sneaked through a first class editing and reviewing process - but it did. 

Sorry Science. It seems like either this little rant, enlightenment from another side, or my hyping Science over Nature has led to correct this paper: In the online version, "ka" is consequently used now (although they eventually explain it with kiloyears ago). We are very happy for the correction and proud, if this website helped to stimulate it. Please send me an email if you cannot find the wrong usage of the units in the listed papers anymore, so I try to include the corrections.

Interestingly, that very same year the very same journal forced one of the major advocates for using "ka" and nothing else as a unit for 1000 years, Paul Renne, to use "ky" which is the favorite unit of  Science for geo-related papers. See here how to persuade authors of journal units.

Here is a recent article on the proper use of time units in Pure and Applied Chemistry
The problem is much deeper in that the second is defined as a SI unit but the year is not, and for example the solar year changes in duration with time as this paper on time unit confusion in New Scientist nicely points out. See also the Wikipedia page on "year". ISO 80000-3:2006 used to define "a" as symbol for the time unit year, although interestingly it defines its length to be either 365 or 366 days (what kind of definition is that ??). The newer version of ISO8000-3:2019 avoids using the year as a time unit and only uses seconds as there is only a precise definition for seconds. NIST also recommends only "a" as a symbol for year. But then NIST refers to the older version of ISO 80000-3:2006 (see above). The IAU recommends "a" as the only unit for year which then is defined as the Julian year which has  precisely  31557600 s when expressed in SI units. The Unified Code for Units of Measure uses subscripts for the tropical year "a_t" ( 31556925.22 s) Julian year "a_j" and Gregorian year "a_g" (31556952 s), but assigns the default year symbol "a" to the Julian year. The IUPAC currently lists year with the unit "a" which stands for the Gregorian year but which is not constant (which probably means that the duration stated is an average value). This diverts from a previous definition given in a joint IUPAC-IUCG paper by Holden et al.(2011) which defines the length of a year as 31 556 925.445 s (derived from the duration of the tropical year in 2000 CE).
Here is a link to an archived Time Unit Discussion page (2009) within the Geological Society of America (GSA) (- the original page has been lost in the recent redesign of GSA pages),   a paper by Aubry et al. (2009) , a paper by Christie-Blick (2011), and  the paper by Holden et al.(2011)- for the PFastu (the Peoples Front for annus as a single time unit) discussing this matter.

Most geoscientists like to have their scales get older from left to right, since they record their data in ages rather than in time.

All time units that include "ago" or "BP" or just the mysterious "a" in "kya" turn time units into negative time units, without making the numbers negative: So the scale gets older from left to right as the numbers count up.

Alternatively people label the scale "age [ka]" instead of "time [ka ago]" so its clear that increasing numbers mean you go back in time. This has the main reason that paleo-records are often from sediment cores, where depth corresponds to an age and usually age increases with depth.

In the paleo-community only modellers tend to do time scales, where ages are negative time and today or the future is on the right hand end of the scale. For all other paleo-people it seems impossible to think of age as negative time as you would do easily with length. Nobody needs a separate unit for length, just because it extends to the left of the coordinate system.

And it seems impossible to use the correct Latin word "annus" for year, as it sounds too similar to "anus" which is used as  the medical expression for the opening of the fecal canal in reference to the terminal cylindrical sphincter muscle (thanks, Paul to clarify on this !). Instead "annum" is used, which sounds very Latin but it actually stands for the duration "for one year" often used in "per annum" instead of "annual" or "per year".

So "a" might stand for the Latin "annus" for year or the plural "anni" for years.
"ka" stands for kiloanni which is thousand years. ok so far.. but then:

It is quite common that "ka" is used for "kiloyears ago", so the "a" somehow stands for both, "annus" and "ago" (??!!) and in contrast, "kyr" is used for a duration (which is always positive) or for a time in the future that would be positive anyway. Reimer et al., Radiocarbon (2009) used a unit, that elegantly solves the problem by simply not using any symbol for year: "cal kBP"  which stands for "calibrated kilosomething before AD 1950". It seems like this was not a mistake, but the radiocarbon community likes "cal kBP" which is heavily used in the description of Marine20, the new marine radiocarbon age calibration curve (Heaton et al., Radiocarbon 2020). And btw - "kB" also stands for kilobyte. And "cal" obviously stands for calories which used to be the unit for energy or heat until it was replaced by "Joule" in 1948. It is also quite common to write "cal yr" for "calendar years" whatever that means: Which calendar do you refer to??

In the end, it is mostly accepted (although not precise and not backed by standards) to use "kyr" for time periods or durations and for time in the future and "ka" for age (or negative time) thus a period from 10.7 ka to 9.7 ka would be assumed to be in the past and have a duration of 1 kyr. Yet, most people use "kyr" and "ka" almost as synonyms and often two or three different ways of writing thousand years are used in the very same paper. And it seems to be perfectly normal to use different units in the text and in the graphs. Most journals have their favorite way of expressing age and time, but as you can see from the list above, this is not consistent and it changes with the field you are publishing in.

And: It is quite common to almost the rule to mix "kyr" for thousand years and "Ma" for million years, since "My" or even worse "my" looks like the possessive pronoun "my", it might be mixed up with the greek letter "µ", and "m" actually stands for "milli" according to SI conventions for multipliers-  Million or Mega have to be a capital "M". Thus "my" would be milliyears (or 8.766 hours). But then "Myr ago" is quite common. Talking about Gigayears, "Ga" might also look like the chemical element symbol for Gallium, while "Gy" is taken for Gray, a unit for the absorbed dose of ionizing radiation. And to complete confusion possibilities, "GY" is eventually used for galactic year, which is actually equivalent to 250 Ma and is based on the time that the sun takes to revolve once around the center of the Milky Way galaxy (which took more like 225 Ma for the last round).

The second very confusing way to use "a" in time/age  units is to explicitly abbreviate "ago" which turns time into age :
"kya" or "k.y.a." or  "KYA"  all mean "kiloyears ago". Its also not rare to ignore SI prefixes and use "tya" for "thousand years ago". There is also an all uppercase version "TYA" which would be "Tera Yotta Amperes" according to SI.
Luckily, as far as we know, nobody has used "kaa" for "kiloanni ago" - but we are sure this will happen soon.
In a 2014 nature communications paper, Grant et al. used "cy" not for centiyears (= 3.65 days) which would be weird but at least the prefix would be correct but for "century" - "cy" why not - who cares about conventions... so - logically,  millennium should be abbreviated "mm" or "ma" for millennia?? here we are... ?? where were we ??

Confused ? feel free to mail us more recent or new examples of additional ways to express thousand years in a yet more confusing way...

Even more confusing: Strictly spoken, "ka" for kiloyears might be ambiguous, since SI still officially allows to use "a" for "are" which was declared the base area unit in 1792 and it means deca meters squared (1 a = 1 dam² = 100 m²). To my personal understanding both the (only) two letter multiplier prefix "da" for "deca" or 10² and the old unit "are" are quite confusing. I think we can live with m² and maybe without the multiplier "deca" if we have to abbreviate it "da". And the chance that somebody wants to use "ka" for kilo ares which would then be 10⁵ m² is rather low. But then nobody would use "Kelvin Yotta Ampere" in an archeological paper (see above)... And "a" also stands for the multiplier "atto" or 10^-18, but "ka" is clear as there is only one multiplier in one unit: kilo-atto does not really make sense - you would just use "femto". If you wanted to use multipliers hecto (you would probably use century - not ha) or - more likely Peta with annus this might be mixed up with the old area unit hectare (ha) or the pressure unit Pascal (Pa)

Conclusion: We still recommend "ka" but whatever you use, the copy editors will teach you, which unit to use in the journal and in the field you are trying to publish in.

A totally different discussion is worth another page: what exactly is "ago" or before present? In the radiocarbon community before present is defined as before 1950 CE (which opens another discussion page on the use of AD, CE, BC BCE ... which literally quickly gets religious - see below).

Carolin et al. (2019) just introduced "ka b1950" which clarifies the reference point, yet it is neither NIST nor SI  compatible. BTW - the unit "b" is used in nuclear physics for "barn" which is 10^-28 m² (or 100 fm²) although not being an SI unit. And as Andy pointed out, it starts to get common to refer to "years before 2000 CE" as "b2k". That is short, it is quite obvious (in the community) - but it is sloppy: like in Reimer's "cal kBP" there is no symbol for year in there! Could be "below 2 km" as well. My suggestion would be to use ka BP₁₉₅₀ instead if necessary (see below).

The expression "ago" is usually less precise. While it is probably irrelevant, whether 500 ka ago refers to 1950 AD or today (what exactly is today ? 2019 CE ?, 1950 AD ??) as reference point, there is a conceptual problem, whether or not we can agree on a fixed point in time that we refer to (like most of us refer to 0°C or 0 K with temperatures) - whether it is pre-industrial, pre-nuclear-bomb or pre-a-currently-popular-religion-protagonist's-wrong-birthdate), since "ago" would otherwise only refer to the time a paper was written.

My personal documentum anni 2019 is Watanabe et al. (2019) in Geology where already in the abstract I found:
"kyr B.P. (where present is A.D. 1950)". This is weird in many ways:
a) why use the restricted number of words of an abstract to start explaining what present in B.P. means?
b) If you assume people know you are not talking about a well known oil company, you might assume that the same people know what BP actually refers to. If you don't,  the "P." in "B.P." should be explained first.
c) So if you decide you want to do this, you could write: " ka BP (kiloyears before present where present refers to 1950 CE) ", correct but horribly cumbersome to be used in an abstract.
d) B.P. is wrong - its BP - you would not use k.m. instead of km, likewise it is AD not A.D.
e) "present is A.D.1950" is wrong because present is not AD 1950  (it was present in 1950 CE), but the radiocarbon dating community agrees to refer to 1950 CE as present.
f)  to be religiously neutral AD (="Anno Domini - in the year of the Lord") is out and CE is in
g) Yet, CE is actually not really free of religious bias since the common era (CE) is defined to start some time around the birth of Jesus Christ which is believed to be at 4±2 BCE - but at least you do not refer to "Lord" or "Christ" in the unit name.
The real advantage is that using BCE makes the sentence "Jesus Christ was born 4±2 BCE" (= between the year 6 and 2 before the Common Era)" sound much less stupid than "Jesus Christ was born 4±2 BC" (= between the year 6 and 2 before Christ ??!!).

My personal documentum anni 2021 is Reinig et al (Nature 2021) which is very nice as it deals with precisely re-dating the Laacher See eruption (LSE) in Germany to 13,006 ± 9 cal a BP. This paper outside of the competition, as the authors only use "years" and not "kiloyears" as time units. Again, BP is being explained in the abstract - yet in a slightly wrong way: "that firmly date the LSE to 13,006 ± 9 calibrated years before present (BP; taken as AD 1950)" - hmmm - the P  in BP is defined as AD 1950, not BP. Then they continue with "calibrated years BP" (long but ok) to then reduce to "BP" without a symbol for year (strictly wrong but quite common - equivalent to AD or CE) to then use the radiocarbon community favourite non-unit "cal. BP" (I would use"cal a BP" or "cal yr BP" if editors insist, although "cal" stands also for calories - see above). To then twist it in figs 3 and 4 to "Time (years cal. BP)" and explain it in the figure caption with "on the calibrated ¹⁴C timescale (cal. BP (AD 1950)). We know what they mean. Should be  "Time (cal years BP)" or "Time (cal a BP)". In this paper, there are indices with BP.  But these indices are used to specify the timescale (which is always relative to "present" which always is 1950 CE). This is smart and compact, yet the index should be used with time: e.g. "Time_GICC05 (cal a BP)" where GCICC05 stands for the "Greenland Ice Core Chronology 2005" timescale.

If you feel you have to explain what present in BP is, you could write BP₁₉₅₀ or BP₂₀₀₀ or BP_2k, which I think is self-explanatory and a bit more compact, plus the index actually relates to the P in BP, so it makes sense to put it right next to it. Or even more compact: use 2k instead of P in BP thus "ka B2k". I would keep a capital B for "before" as it is established. This is my suggestion as it has not been used anywhere to my knowledge. But it is mostly obsolete: ka BP is fine as long as the community sticks with 1950 CE as a reference point (as in civilized countries temperatures are written in degrees Celsius or relative to 273.15 Kelvin without further explanation).

My personal documentum anni 2022 has just become Köhler et al. (Paleoc Paleoclim 2022) where instead of "cal ka BP" the authors use "kcal BP" throughout the paper which is consistent with notoriously using "cal BP" instead of "cal years BP" which is hip in in the radiocarbon community but even worse than "just" omitting any unit for year by using "cal kBP" (Reimer et al.Radiocarb. 2009, see above): Everybody who ever had to worry about his weight knows that "kcal" - though not a SI unit - is widely used for kilocalories and it is equivalent to 4.1855 kJ or the amount of energy needed to increase the temperature of 1 L of water by 1°C (or by 1.8°F for those of you from the remote part of the world where centigrades sound exotic -> see above).

Getting deeply into dating and timescales is worth another few webpages and not the scope of this one. When it comes to e.g. radiocarbon dating samples, the residual radiocarbon activity is converted to a calendar age with calibration relations that change over time. Calendar ages differ from radiocarbon ages because radiocarbon ages are calculated assuming a constant initial radiocarbon activity. As the radiocarbon activity has changed naturally and anthropogenically over time this has to be accouted for. IntCal20, SHCal13 and Marine20 are recent versions of those calibrations that were done separately for different parts of the world. These calibration relations are constructed using natural archives like tree rings, marine and lacustrine sediments, corals and speleotherms, which have both, a radiocarbon signal and annual structures that can ultimately be counted.

Increasing numbers of snippets of annually resolved records are being spliced to give increasingly well justified conversions of residual radiocarbon activity to calibrated calendar age. So choose if you take "cal" as calendar or calibrated or both. The year count in these conversions is initially based on the number of times Earth revolved around the Sun causing the patterns in the annual banding that are being analyzed (regardless of the precise duration of one revolution). This is why the community speaks about calendar years which involve a full seasonal cycle but are initially less defined in their precise duration. Since records with annual structures may only be spliced when the seasonal patterns are comparable, those calibration conversions have been prepared  for the Northern and Southern hemispheres and for marine records separately. More importantly, the result of this conversion is never a defined point in time, but a time interval which represents a probability range. And these dated age ranges refer to a certain timescale that might be revised and replaced by more precise conversions with time. So timescales are changing with time.

Outside the competition but worth noting: Garcin et al. (Nature 2022) - a nice paper by the way. Again, the journal made the authors explain already in the abstract what present is: "17,500 calibrated years before present (cal. yr BP; taken as AD 1950)". Mostly similar to mistakes as in the 2019 Paper of  Watanabe et al. (2019)  and in  Reinig et al (Nature 2021) above, looks like Nature loves the expression "taken as AD1950" without making clear what this refers to.
a) its "cal" not "cal.",  b) why explain BP in the abstract, c) why explain BP at all, d) The "P" in "BP" is explained not "BP" or the whole unit d) its not AD 1950 its 1950 CE (see above).
My suggestion is to use either "17.5 cal ka BP" or "17.5 cal ka BP₁₉₅₀" to make the reference point clear. Or recalculate to before 2000 CE and use "17.55 cal ka BP_2k" or "17.55 cal ka BP₂₀₀₀" which makes conversions to years CE in archeology a bit more straight forward.

Getting carried away....
Interestingly the "Common Era" was originally called the "Vulgar Era" (annus aerae nostrae vulgaris) when they were hunting witches in Europe as they went from the Medieval Warm Period to the Little Ice Age and the pandemic to fight was the Black Death. They had no clue then, how vulgar exactly politicians would get as witch-hunts and pandemic would re-emerge when mankind was fighting global warming (or not really). So now that we learnt how vulgar people in the most prominent positions may get in the 21st century we might want to switch back to  BVE / VE. We could start to refer to 2016 VE as  0 a BT. 1950 VE would then be 66 a BT. Luckily 2021 will be 1 AT which would leave us with a four year gap. As many of you are probably not comfortable with referring to the election year of an infamous president anyway, we could give it a more positive spin: We could use 2012 - the year the model S was introduced - as the beginning of the ECE Electric Car Era so 1950 CE would then become 62 a BT (before Tesla). This is my recommendation as Tesla is an SI derived unit already ......... Just kidding... 

If you are confident that fighting global warming will succeed, we could also choose 2018 as a reference, the year when a collapse of CO₂ emissions began as "Greata" started spending Fridays on streets. So 1950 VE would then translate to 68 a BG (before Greta). But then "G" is already used as the gravitational constant (6.674×10^-11 m³ kg^-1 s^-2) and for the multiplier prefix Giga or 10⁹. Yet, I personally would prefer BT (before Tesla). If the global perturbations get as severe as it currently looks, it might also make sense to complete confusion and redefine BC in a virological rather than in a theological way and refer to the year 2020 CE or -70 a BP as 0 BC (Before Corona) so "present" in a radiocarbon sense would become 70 BC and Jesus Christ would have been born in 2024±2 BC. Btw: Has anybody thought about the fact that if we refer to 1950 CE as present, we logically should refer to 2021 CE as 71 a AP (70 years after present).  Stay healthy and make sure there is a future for most of us !!

Getting completely carried away: Because in 1977, John Cleese was writing his groundbreaking Sciense paper about theories for a new programming language (Monty Python), Graham Chapman got into the competition and successfully submitted his own elaboration on 100 ka cycles being nonsense after he had seen those stamp sized frequency analysis plots in Hays, Imbrie & Shackleton (Science, 1976). In early shots for 'Life of Brian', Graham Chapman wanted to incorporate this recent experience of submitting a geo-related manuscript. So John Cleese helped him to give a short demonstration of what happens, if you think its o.k. to select a NIST / SI compatible unit for kiloyears when you want your paper published in Science. We superimposed subtitles that represent the rarely heard dialogue from the lost original script which only in 1979 evolved into the famous Latin lesson scene. How to not get your story on the cover on the Science Mag has been recently explained here (long version) .

We only found out in May 2025 - almost 50 years after the iconic paper of Hays, Imbrie & Shackleton (Science, 1976) was published - that it is also iconic in using "Years x 10³ B.P." as yet another weird unit for age. This marks now after 15000 reads of this page the last entry in our list (currently the 79^th). Hays, Imbrie & Shackleton managed to squeeze evidence for "100 ka" orbital cycles out of a 450 ka composite record that was dated by one radiocarbon date and by matching it with orbital cycles. Interestingly, the four cycles covered by the composite record were (termination to termination) 117 ka, 120 ka, 89 ka, and 89 ka long and a "100 ka" peak only appeared in the stamp size frequency plots, because the authors introduced using a frequency rather than a cycle length scale which was extended all the way to a frequency of 0.167 cycles/ka (which is equivalent to a cycle length of 6ka). This way everything longer than 70ka cycles was nicely squeezed into one sharp peak around 0.01 cycles/ka, which has been interpreted as "100 ka" cyclicity. But the width of the peak - as narrow as it appears - actually covers cycle lengths from 70 ka to about 150ka. Still ever since this paper has been the reason for generations of paloeoceanographers to fantasize about the drivers for "100 ka" cyclicity.

I know any doubts concerning the transition from obliquity driven 41ka to ??eccentricity driven?? "100 ka" cycles in our community comes close to blasphemy but sorry - I identify as a FOO fighter (where FOO stands for friends of obliquity) because as far as I can see there is hardly any glacial cycle that has a length of 100 ka, unless you include an error margin of > 20 ka.

One more thing: This website has already been cited as a reference for time units. If you want to use this webpage as a reference please use 2015 as a publishing year because that is the year it first went online. As you can see below, the site is constantly updated, so it makes sense to include the latest time stamp. 

Kölling, Martin (2015): Numerous ways to say "thousand years" in a scientific paper.
Version May 6, 2025 - www.sedgeochem.uni-bremen.de/kiloyears.html

,
You might also want to refer to the more serious papers of  Christie-Blick (2011) and  Aubry et al. (2009).
I personally recommend to join the "Popular front for annus as single time unit (PFastu)" and cite Holden et al.(2011).

If this page helped you, you can help us by looking at our paper in nature geoscience
It is on CO₂ release from continental shelves and you can read it for free here

Even if you are not interested in the shelf life of pyrite, the paper has a wonderful graph
that shows the relation between CO₂ and sea-level during the last 800 ka.
It shows impressively how crazy our current CO₂ levels are.

Martin Kölling, MARUM

Btw: I am a porewater geochemist, not a radiocarbon person or dating specialist. So if I am getting things wrong, please feel free to tell me.

One last thing: If you think I am what even educated Dutch would very metaphorically call a "mierenneuker" you have not seen any piece of tidied up art by Ursus Wehrli.

original time stamp 23.July 2015
updated 29.Aug15, 30.Aug15, 24.Sep15, 1.Oct15,  2.Feb16, 14.Sep16, 15.Dec16, 30.Mar17, 13.Jun17, 25.Oct17, 12.Nov17, 4.May18, 7.June18, 6.July18, 7.Aug18, 17.Dec18, 9.Jan19, 16.Jan19, 20.Mar19, 27.Mar19, 27.Jun19, 6.Sep19, 16.Oct19, 20.Oct19, 23.Oct19, 7.Nov19, 12.Nov19, 1.Apr20, 2.Apr20, 5.Apr20, 17.Jul20, 10.Sep20, 24.Oct20, 20.Jan21, 30.Mar21, 10.Apr21, 4.Jul21, 5.Jul21, 3.Sep21, 17.Sep21, 22.Sep21, 8.Oct21, 22.Apr22, 27.Oct22, 3.Nov22, 7.Feb23, 19.Apr23, 26.Apr.23, 28.Apr23, 15.Jan25, 26.Feb25, 6.May25

Thanks to Paul Renne, Michael Deckers and many others for comments and improvements and to all the "contributors" for the creativity in finding dozens of wrong ways to express the unit kiloanni.