One of the key pieces of evidence in the case of the Star of Bethlehem is the record of the object seen in March 5BC, classed by Clark and Stephenson in 1975 as a probable nova on the basis of its long duration and fixed position. People also make great store of the evidence that the Chinese records are seriously incomplete between 20 BC and 1 BC because there is a shortage of both comets and novae in the Chinese chronicles. Similarly, people look at the long interval since the last bright nova was observed and start to wonder where the bright novae have gone.

If we take the 20^th Century as a guide, bright novae have been extremely common, so we would expect them to have been just as common two thousand years ago. The list of novae that reached at least magnitude +2 in the 20^th Century is impressive, especially when we consider that the 19^th Century closed with a bright nova in 1898:

Name	Year	Magnitude
GK Per	1901	0.0
V.603 Aql	1918	-1.1
V.476 Cyg	1920	2.0
RR Pic	1925	1.1
DQ Her	1934	1.2
CP Lac	1936	1.9
CP Pup	1942	0.4
V.1500 Cyg	1975	1.8

That makes an average of one nova of magnitude +2 every ten years over eight decades, with an extraordinary six of them appearing in twenty-four years between 1918 and 1942.

There are just fifty stars in the sky that are magnitude +2.00 or brighter – exactly fifty, in fact – thus, we may assume, quite reasonably, that any new object of magnitude 2 or brighter should be unmistakeable to anyone with a good knowledge of the sky. Proof of this is that when Nova Cygni blazed out on August 29^th 1975, it was discovered by Osada, in Japan and, at nightfall in Europe, literally hundreds of people discovered it independently. To my enormous chagrin, I missed observing the nova myself: it appeared on a Friday evening; I only heard about it on Sunday night and, by the time that I had a chance to look for the nova myself, on Monday, it was already as faint as magnitude 5 and fading fast.

Since 1975 though, if we take magnitude +2 as the peak magnitude that defines what is a bright nova, there have been no bright novae in what is now more than forty-two years. Of the 239 novae discovered since Nova Cygni 1975, the brightest have been: Nova Centauri 2013, at magnitude 3.3 (in the far south of the sky); Nova Scorpii 2007/1, at magnitude 3.9; Nova Cygni 1992, at magnitude 4.2 and Nova Sagittarii 2015/2, at magnitude 4.6. All were naked-eye visible, but none were bright enough to be unmissable in the way that the objects listed in the table above were.

There is a feeling among observers that the sky owes us a bright nova. However, we can also look further back into the records of novae since 1600 and what is generally accepted to be the dawn of modern astronomy and attempt to get a better idea of the statistics. Maybe the 20^th Century has been a statistical anomaly and bright novae are not as common as they appear. Of course, this modern period started in spectacular fashion with Kepler’s Supernova in 1604. Between 1600 and 1900 there are just thirty-nine reports of novae or nova-like objects[1], all but five of them posterior to 1840. That though is to be expected, as techniques became more sophisticated, fainter novae were being detected more systematically.

However, the distribution still looks non-random. Only one nova, was reported in the entire 18^th Century and that was the magnitude 6, Nova Sagittae 1783. In contrast, there were two novae of magnitude 3 or brighter reported in the 17^th Century (Nova Vulpeculae 1670 and Nova Puppis 1673, although the former is no longer regarded as a classical nova) and three between 1848 and 1900 (Nova Ophiuchii 1848, Nova Cygni 1876 and Nova Sagittarii 1898).

Whatever way you look at the numbers though, there were far fewer bright novae reported between 1600 and 1850 than between 1850 and 2000. And the period between 1898 and 1975 contains all but one of the bright novae reported since 1600.

There are two obvious possible explanations for this:

• The peak in the frequency of appearance of bright novae between 1898 and 1975 is a genuine statistical anomaly and that bright novae are, on average over a long period, less common.
• The peak in the frequency of appearance of bright novae between 1898 and 1975 is due to statistical bias in discoveries.

In the former case, it is a real effect, that appears for a unknown reason. Maybe it is just that we were extraordinarily fortunate to see so many bright novae in such a short period of time. Sometimes you will just get lucky tossing a coin and toss ten heads in a row but, the longer that you continue to toss the coin, the more the number of heads and tails will even out. In other words, any statistical series will come twice as close to the long-term mean over four centuries as it does over one, if we are treating pure Poisson statistics (a purely random phenomenon).

This effect is well known to meteor observers (and to users of London buses). Poisson statistics have the effect that meteors appear to appear in bunches, separated by long intervals: even around Perseid maximum you can have periods of five, ten or even fifteen minutes when not a single meteor appears and then three or four bright ones appear almost simultaneously – the average ends up being close to one-a-minute, but there are statistically more intervals shorter than an minute than longer ones and so the impression is that the meteors appear grouped-together in a totally non-random fashion.

In the latter case though, we would have to assume that the novae were appearing, but were not being detected for some reason.

There is one obvious reason why nova discoveries were much less frequent prior to 1850 and that was the lack of systematic observing of the southern sky. Of the bright novae of the 20^th Century listed above, one could not have been discovered from the northern hemisphere and another would have been difficult to observe except from the extreme south of Europe. We can suggest that an entire third of the sky was little-observed prior to the mid-19^th Century and that novae in that third of the sky would not have been discovered: that reduces the anomaly a little.

Is it reasonable to argue that the rest of the anomaly is down to observers just not noticing when a nova appeared?

The 17^th, 18^th and early 19^th centuries featured a large number of renowned visual observers. Apart from the astronomers who mapped the sky and added new constellations and those who produced increasingly voluminous star catalogues, many assiduous observers such as Messier, Méchaine, Pons and William Herschel were active. The large number of increasingly faint comet discoveries is evidence that there was intense scrutiny of the sky and that anomalies were being picked up, although it is true that nebulous objects stood out to these observers more than stars, as evidenced in turn by the fact that it was not until a systematic search campaign was initiated, that even the brightest asteroids were discovered. However, it was also a time when everyone knew the constellations as well as their own neighbourhood and you would think that a nova of magnitude +2 would be noticed immediately by hundreds of people and commented.

As such, even if a significant fraction of bright northern hemisphere novae were simply missed, one has to suspect that the frequency of bright novae must have been lower than it was at the peak in the 20^th Century.

We can also get a hint at the frequency of bright novae in the past from Chinese records. Table 3.1 of the Clark and Stephenson book, “The Historical Supernovae” lists their seventy-five candidate novae and supernovae of the pre-telescopic era from Oriental records. They class the reliability of identification of objects as a nova or supernova on a 5-point scale from 1 to 5, with objects of class 1 and 2 likely or certain novae of supernovae and classes 4 and 5 extremely tentative candidates. Astonishingly, for a record that covers two thousand years, there are just twenty objects that were of long duration and high reliability, nine of which are now known to have been supernovae. The 5BC Star of Bethlehem candidate is the only high-reliability nova/supernova candidate object in BC dates. Granted that in choosing long duration of visibility as a criterion Clark and Stephenson were biasing their statistics against fast novae but, even if we eliminate the period between 500BC and 1BC for which we know many records were lost, one strong nova candidate every century and a half seems an astonishingly small number. As a result, we might suggest that the lack of records is simply a statement that the Chinese were only interested in K’o-hsing – Guest Stars – if they were particularly brilliant: it seems likely that anything that was not negative magnitude was not of any great interest, while the very brightest and longest-enduring objects were described in great detail.

So, we come back to the original question: where are the bright novae now and where were they in the past?

My guess is that:

• The high frequency of bright novae between 1898 and 1975 was a statistical anomaly. Rather than 10 per century, the true rate is more likely to be 3-4, consistent with the observed rate through the 19^th Century and, certainly, lower than 5. Evidence for this is the fact that there has been just one bright nova in the 76 years since 1942, consistent too with the rate of bright novae in the 19^th Century.
• Part of the deficit of novae in the 17^th and 18^th centuries can be explained due to lack of coverage of the southern hemisphere. If random chance deemed that several consecutive bright novae happened to fall in the southern hemisphere (i.e. similar to obtaining 3 or 4 consecutive heads when tossing a coin) a century could pass without a single, bright nova being observed.
• The lack of recorded bright, Chinese novae can be attributed to a combination of bright novae being less common than recent statistics have led us to believe, to the Chinese only recording K’o-hsing that were particularly bright and to the selection effect of only regarding objects of long-duration as candidate novae, thus eliminating most fast novae from the statistics.

If so, the fact that we have not seen a bright nova since Nova Cygni 1975 is not a statistical anomaly: it is a reflection that the statistics are normalising and that, like London buses, we must be patient and, suddenly, three will come along at once. In this case, long-term, observing a bright nova may actually be a once or twice in a lifetime event, rather than a regular occurrence and those people who observed so many bright novae in the first half of the 20^th Century were extraordinarily fortunate. It also makes the 5BC event, if it was a nova, a far more significant object than we have believed it to be thus far.

[1]This includes the outburst of the recurrent nova T Coronae from 1866 and Eta Carinae in 1843, which are frequently excluded from lists as neither is a classic nova. I propose not to treat either as a genuine nova.