Q. I prefer to use the spelling "Hipparchos", or sometimes "Hipparcus" when describing the Hipparcos mission or the Hipparcos Catalogue. I feel this underlines better my classical education. Is this ok?

A. Sorry, you are incorrect...

A. Hipparcos is an acronym for HIgh Precision PARallax COllecting Satellite. Appropriately the proununciation is also very close to Hipparchus, the name of a Greek astronomer who lived from 190 to 120 BC. By measuring the position of the Moon against the stars, Hipparchus was able to determine the Moon's parallax and thus its distance from the Earth. He also made the first accurate star map which lead to the discovery, when compared with other data from his predecessors, that the Earth's poles rotate in the sky, a phenomenon referred to as the precession of the equinoxes.

The concept of using the data recorded by the star mappers for astrometric and photometric observations was conceived by Erik Høg, a Danish astronomer involved with the Hipparcos mission. It was fitting that the catalogue which resulted from the star mappers should then be named after Tycho Brahe, a 16th century Danish astronomer, who produced the first 'modern' star catalogue (1602).

Q. Please explain the catalogues in the context of some other recent catalogues that I've heard about which make use of the Hipparcos and Tycho data, in particular the ACT and the TRC? [Aug 99]

Details provided by Ronnie Hoogerwerf, Sterrewacht Leiden (August 1999)
At the end of the 19th century photographic plates, for the first time, could be produced and used in a systematic and reliable way. This led the astronomical community to propose the `Carte du Ciel' project: an enormous international effort to `map the heavens' (see, e.g., Eichhorn 1974; Debarbat et al. 1988; Urban & Corbin 1998). The main goal was to make an astrometic catalogue consisting of all stars brighter than about photographic magnitude m=11 mag with positional accuracies better than 0.5 arcsec (this part of the project was called the Astrographic Catalogue). The second goal was to produce charts with the relative positions of all stars down to 14 magnitude (the Carte du Ciel). This second part of the project was never completed due to the enormous amount of time and costs needed to reproduce and distribute these plates. The first project was completed. The entire sky was imaged on photographic plates. To reduce problems with plate defects, each sky area was observed twice by overlapping plates. This major undertaking was distributed over 20 observatories, each of which observed part of the sky (subdivided according to Declination). The observatories used the same type of telescope to ensure that the plate scale would be identical for all plates. In total, 22,660 plates were exposed. The plates were developed, measured, and the stellar positions recorded (in millimeters), along with the known plate corrections and the equations needed to convert millimeters into Right Ascension and Declination. Typographical errors in these publications were published separately. These publications together with their corrections are known as the Astrographic Catalogue (AC).

Tycho: The Tycho instrument functioned as the star mapper for the Hipparcos satellite: it recorded all stars passing the field of view. These measurements were reduced to obtain positions, proper motions, parallaxes, and broad band photometry for 1,058,332 stars. The errors in the astrometric parameters (about 20 mas/yr) are typically 20 times the errors in the Hipparcos Catalogue. However, the TYC positions are still better than in any other catalogue of comparable size. A re-analysis of the raw data is currently being carried out and will result in a new Tycho Catalogue (TYC2) (Hoeg 1997) containing about 2.5 million stars; it will become publicly available in the first half of the year 2000.

AC2000 & ACT: The AC2000 (Astrographic Catalogue 2000) (Urban et al. 1998b) is the result of a new reduction of the AC conducted at the USNO. Magnitude and positions are provided for 4,621,836 stars in the Hipparcos reference system (ICRS, see ESA 1997, Vol. 1, Section 1.2), with a median epoch of 1904. The AC2000 has been combined with TYC (median epoch J1991.25) to obtain proper motions based on the position difference between both catalogues; the average epoch difference is about 90 years. These proper motions are recorded in the ACT (Urban et al. 1998a). The TYC stars were matched to the AC2000 stars by propagating the Tycho positions to the AC2000 epoch, taking into account the TYC proper motions. An area of 15 arcsec radius was then searched around the propagated TYC position for AC2000 stars (see Urban et al. 1998 for details). In total 97.5 per cent TYC stars were unambiguously matched to an AC2000 entry. Some TYC stars were considered problematic and discarded if they were indicated to be part of close multiple systems, or if the positions had extremely low astrometric quality. After the cross-identification, proper motions were computed using the differences in position and epoch. The proper motions were brought from the mean epoch of the AC2000 and TYC to the epoch J2000 (Fricke 1963). In total, the ACT contains positions and proper motions for 988,758 stars (average error in the proper motion of 2.4 mas/yr and better than 3.5 mas/yr for 80 per cent of the stars, positional accuracies adopted from the Tycho Catalogue).

TRC: The TRC is the European equivalent of the ACT. Its construction also required a re-reduction of the AC, putting the AC onto the Hipparcos reference system, and combining the resulting position with those in the TYC. The TYC stars were propagated to the AC epoch without using proper-motion information; the relatively large proper-motion errors might degrade the stellar position instead of improving it. The propagated TYC positions were then matched to stars in the AC using a maximum search window of 180 mas/yr × dT, where dT is the epoch difference between the TYC and the AC. Consequently, high proper motion stars are not included in the TRC. Doubtful identifications were, if possible, resolved using magnitude information in the two catalogues. Stars with uncertain photometry or astrometry have not been included in the TRC. The TRC proper motions were constructed using the TYC and AC positions. On top of the random error an additional 1.6 mas/yr was added (in quadrature) to account for residual systematic effects. The TRC contains 990,182 stars (average error in proper motion of 2.4 mas/yr, positional accuracy of 25-45 mas). Details of the construction of the TRC are given in Kuzmin et al. (1999).

References:
Debarbat S., Eddy J.A., Eichhorn H.K., Upgren A.R., 1988, Mapping the Sky: Past Heritage and Future Directions, IAU Symp. 133
Eichhorn H.K., 1974, Astronomy of Star Positions. Frederick Ungar, New York
Hoeg E., 1997, ESA SP-402, 25
Kuzmin A., Hoeg E., Bastian U., Fabricius C., Kuimov K., Lindegren L., Makarov V.V., Roeser S., 1999, A&AS, 136, 491
Urban S.E., Corbin T.E., 1998, Sky & Telescope, Vol. 95, No. 6, p. 40
Urban S.E., Corbin T.E., Wycoff G.L., 1998a, AJ, 115, 2161
Urban S.E., Corbin T.E., Wycoff G.L., Martin J.C., Jackson E.S., Zacharias M.I., Hall D.M., 1998b, AJ, 115, 1212

Q. Why doesn't the Hipparcos www site allow interrogation according to various important cross-identifiers: BD, or CD for example? [Dec 97]

A. Cross identification of stars is frequently complicated by the angular precision or resolution of earlier catalogues, or errors propagated in these identifiers through the literature. Some identifiers are included within the fields of the machine-readable version of the Hipparcos and Tycho Catalogues, but they should be considered more "for information". If you know a particular star identifier, use the CDS SIMBAD facility to find the corresponding HIP number - these identifiers are maintained and updated. Or use the position of the star to search the catalogue.

Q. I've found a couple of cases where the cross-reference given for a HIP star is different to that given for the corresponding TYC star. What is going on? [Dec 97]

A. This is a good example of the previous problem. The HIP and TYC catalogues were cross-referenced to other catalogues independently. Sometimes, there is no obvious answer as to which is "correct". See SIMBAD!

Q. If I know a TYC identifier, I can find its corresponding HIP number (if such a correspondence exists) through Field T31. If I know only a HIP number, how can I find its corresponding TYC entry? Why don't you provide a cross reference that way round?

A. This could be done, but things get a little messy due to the different angular resolutions of the two catalogues, and the fact that a HIP number may embrace a double or triple system, for which the components are separately listed in the Tycho Catalogue. We suggest you proceed as follows: use the HIP number to find the position of the object. Search at this position, with a radius suitable to include multiple components (say 30 arcsec). All relevant TYC objects will be returned.

[In the Celestia 2000 CD-ROM, the HIP window provides the TYC number of any TYC entry in common with the HIC entry, and vice versa.]

Q. Concerning binaries and the effect on parallaxes, L. Szabados has reported (Hipparcos Venice '97, ESA SP-402, page 657, available at this site) that undetected orbital motion in Cepheid binaries "falsifies the trigonometric parallax determined from the Hipparcos measurements". What does the project have to say about that? Are all the Hipparcos parallaxes suspect? [Jan 98]

A. It may be worth you taking a look at Volume 3 of the published catalogue, ESA SP-1200, describing the catalogue construction, and especially Chapter 20 "Verification of Parallaxes" and Chapter 22 "Analysis of Double Star Results". Basically, we do not expect the parallaxes to be significantly perturbed by (detected or undetected) binarity, unless the orbital period is close to one year, in which case the photocentric motion may confuse the shift in the system's apparent position due to parallax. ASCII Discs 5 and 6 (in Volume 17) provide the intermediate astrometric data necessary to carry out a rigorous re-analysis of the astrometric data, taking into account subsequent ground-based information on orbital parameters of specific objects. To our knowledge this has not been carried out for these objects. So some suspicion is always appropriate, but we are not aware of specific errors confirmed by a thorough analysis.

Q. In a recent paper accepted in MNRAS "The Lutz-Kelker bias in trigonometric parallaxes" by R.D. Oudmaijer et al, they state in the Abstract "This statistical effect, the so-called Lutz-Kelker bias, causes measured parallaxes to be too large." And later "We find that there is indeed a large bias affecting parallaxes, with an average and scatter comparable to predictions." Does this mean that all the Hipparcos parallaxes are biased? [Jan 98]

A. Yes and no. Not in the simple sense that there is some "mistake" in the catalogue that with hindsight could or should have been fixed. Lutz-Kelker bias, and the bias in quantities *derived* from the measured parallaxes, are well-known effects which are not disputed. These effects should be considered in any rigorous discussion of derived quantities, see, for example, two papers in the Hipparcos-Venice '97 Proceedings, ESA-SP 402 (available at this site): Luri & Arenou "Utilisation of Hipparcos data for distance determinations: error, bias and estimation", page 449; and Brown et al "Some considerations in making full use of the Hipparcos Catalogue", page 63. But there are several issues involved, and these further comments may be consulted for additional details.

Q. The distance scale determined from the Hipparcos Cepheid measurements by Feast & Catchpole (1997, MNRAS, 286, L1) gives a distance modulus of the LMC of 18.70+/-0.10. Now others (e.g. Oudmaijer et al.) are revisiting the results using a "proper Lutz-Kelker correction". Who is correct? [Jan 98]

A. Feast & Catchpole specifically formulated the problem to avoid biases in the luminosities. We can only leave it to the refereed literature to make the position clearer in due course.

Q. A preprint by Pinsonneault et al suggests that the Hipparcos parallaxes might have a systematic error of 1 mas in some areas of the sky, particularly for the Pleiades, which would explain some main sequence fitting discrepancies. Is this possible? [Jan 98]

A. The Hipparcos observations and data analysis was carried out in such a way that any systematic parallax shifts should be common to all stars. There is good evidence that a *global* shift does not seem to be present [see for example Volume 3 of the printed catalogue, Chapter 20 on verification of the parallaxes]. The Hipparcos Catalogue documentation does not address explicitly possible systematic offsets occurring at small angular scales, although the general effects of correlations on such scales is discussed. What little we know presently is mostly implicit in the discussion of the NDAC-FAST consortium differences given in Chapter 16 and 17 of Volume 3. Hard facts are seemingly: (1) that there is no significant *global* offset; and (2) that the quoted standard errors are not significantly underestimated, again on a global scale. This is what is proven in Chapter 20 and in many subsequent applications of the data. But it still leaves room for spatial correlations - which in a single experimental realisation are seen as systematic errors on small scales. We don't know what the spatial correlations are, but values of up to 0.2-0.3 over a few degrees in bad (ecliptical) regions may not be unrealistic, given the correlations of the FAST/NDAC differences. This would correspond to an expected (rms) level of systematics that is of the order of (1.5 mas)*sqrt(0.25) = 0.75 mas, if the typical error of the individual parallax is 1.5 mas. In Figure 16.34 (Volume 3) it is shown that NDAC/FAST differences could easily be 2 mas when averaged in 2*2 deg fields in the ecliptic region. Even if the merging reduces the systematics by a factor 2 there could still be remaining effect at the 1 mas level in 2*2 deg areas. So the behaviour reported is probably consistent with our knowledge of the catalogue, although whether this is the correct, or complete, explanation here will probably require further investigations. It would be interesting, for example, to look at the separate NDAC and FAST parallaxes for the Pleiades stars. At the same time, the Pleiades is not the only cluster showing results discrepant with pre-Hipparcos predicted parallaxes.

Q. So what evidence is there that the astrometric data for *individual* objects are reliable?

A. Astrometry (positions, proper motions, or parallaxes) having a demonstrated accuracy comparable to that of Hipparcos at the individual object level are strictly limited. This of course makes it very difficult to quantify the Hipparcos errors from an external perspective - a problem common to all experiments. However, VLBI results provide one such source, and these have supplied confidence in the overall positional quality (see ESA SP-1200, Volume 3, Chapter 18, especially Table 18.1), as well as in the individual parallaxes (Chapter 20, Table 20.1 and Figure 20.2). Independent distance estimates have also been obtained for three Hyades orbital binaries, with parallax values (in mas) as follows:

HIP 	Hipparcos	Torres et al 	Reference
20087 	18.25+/-0.82 	17.9+/-0.6   	Ap.J., 1997, 474(256)
20661 	21.47+/-0.97 	21.44+/-0.67	Ap.J., 1997, 479(268) 
20894 	21.89+/-0.83 	21.22+/-0.76	Ap.J., 1997, 485(167) 

Q. How about the existence of comparable quality astrometric data in conflict with the Hipparcos results?

A. One of the most significant tests is presumably the Hubble Space Telescope FGS results on the following Hyades cluster objects, reported by a large collaboration (van Altena et al, Ap.J., 1997, 486, L123). Here are the FGS-based parallaxes (in mas) and proper motions (in mas/yr), compared with the Hipparcos determinations:

HIP      pi(FGS)  pi(HIP)       mu_a(FGS)   mu_a(HIP)      mu_d(FGS)  mu_d(HIP)
20563   15.4/0.9 19.35/1.79    105.0/0.9 112.72/2.48    -14.1/0.8 -33.20/1.99
20850   16.6/1.6 21.29/1.91     78.9/1.3 106.16/1.57    -16.7/1.5 -17.59/1.18
21123   16.5/0.9 23.41/1.65    106.6/1.1 105.81/1.60    -16.2/3.0 -30.97/1.25
21138   15.7/1.2 15.11/4.75    102.5/1.4 100.76/4.54    -14.0/1.4 -23.42/3.21

Q. How do I get more details of objects shown in the Millennium Star Atlas? For example, how do I get the period and other details of the variable objects shown in the Atlas?

A. There are two easy ways. Either use the www on-line catalogue search facility to get details of objects in a given area close to that of the object you're interested in. It should be straightforward to track the object down from that. Or use Celestia 2000, with its interrogation software for PC and compatibles, to perform this search.

Q. Why does the Millennium Star Atlas show distances in light-years, when the rest of the Hipparcos Catalogue works in milliarcsec and its related reciprocal, the parsec?

A. Most of the Hipparcos and Tycho Catalogue material is primarily of use to professional astronomers, who tend to prefer use of the parsec as the unit of distance measurement. The Millennium Star Atlas should serve the professionals, but we hope it will have great value for the amateur astronomer. Sky & Telescope know that from their correspondence, the light-year is more welcomed. We felt the use of the light-year was therefore preferable in this context. If an accurate distance (and error) estimate is needed, the user will have to extract these data from the main catalogue in any case.

Q. There seems to be an error in the Millennium Star Atlas, chart 736. At the bottom right there is a binary with a common proper motion. But one is labelled "51 ly" and the other "44 ly". Is this an error in one of the parallaxes, or is one of the components itself an astrometric binary whose erratic motion has yielded an incorrect parallax? [Jan 98]

A. The two stars are HIP 42762, V=11.83, parallax=64.56+/3.92; and HIP 42748, V=9.62, parallax=74.95+/13.82. The brighter star has an abnormally large error on its parallax. Hipparcos flags it as suspected non-single. It shows the importance of looking at all of the information, including the uncertainties - of course, it was not practical to show these in the Atlas.

Q. What can amateur astronomers contribute to follow-up observations now that the Hipparcos and Tycho Catalogues are available?

A. A huge amount! A large number of new variables have been discovered, and many of these reveal interesting features in their light curves that require follow-up observations. For example, there are many new eclipsing binaries for which only a single minimum has been observed by Hipparcos. These need additional ground-based observations to pin down their period. We refer interested observers to the www pages of the AAVSO (see links from this site) for more details.

Q. I want to order the Hipparcos and Tycho Data on CD-ROMs. What should I get: Volumes 1 and 17 with the data in ASCII form, or Celestia 2000, which I understand has all the data along with interrogation software? [Jan 98]

A. The former is strictly ASCII tabular data. It includes all of the public released Hipparcos and Tycho data, including intermediate astrometric data, solar system objects, epoch photometry, and the documentation and figures in pdf form. Celestia 2000 contains the most important, or most frequently used subsets of the data, with access software specifically for PC compatible platforms. You can do arbitrarily complex searches on the catalogue (e.g. extract all A stars, with magnitudes between 5-7, parallaxes larger than 50 mas, and lying between declinations 50 to 70 degrees. It depends what you want to use the data for. If you are an amateur astronomer, for example, almost certainly Celestia 2000 will be what you need. If you want to import the data into your department's data base, the ASCII CD-ROMs are probably what you want. Some basic access software is provided (written in C and tested on Unix platforms), but generally you will need to do a little more work in extracting the detailed data sets.

Q. Why are there no radial velocities listed in the Hipparcos Catalogue? Isn't it obvious that these are rather closely connected with the other five astrometric parameters, and anyone studying stellar kinematics would like to have them in a convenient form?

A. The Hipparcos and Tycho Catalogues are purely observational - radial velocities were not acquired by the satellite, and attempts to undertake a coordinated radial measurement programme from the ground, such that they could be included in the final catalogue, were unsuccessful. In any case the published catalogues were not intended as an opportunity to create a wider stellar data base. For this, users should resort to the extensive facilities provided by the CDS; in particular updated radial velocities can be found through SIMBAD. [Celestia 2000 also contains a number of radial velocities, which were originally compiled at the time of the Hipparcos Input Catalogue compilation, which can be accessed object-by-object. Celestia 2000 has been significantly updated with respect to the values published in the Hipparcos Input Catalogue (1992), although there are many more recent radial velocity results specifically acquired by observers for the Hipparcos stars which are not included, because their acquisition and reductions are not yet finished (or published). Celetia 2000 also allows major cross-identifiers to be accessed at the same time.]

Q. Griffin et al. (Observatory, 1997, 117, p351) make the following commentary: "We notice that Hipparcos, which had already re-named more than 10**5 well-known stars with its own numbers, did not take cognizance in its Input Catalogue of the duplicity of HR 51565/6, so we may count ourselves lucky that it did not assign yet another redundant designation to it when it re-discovered that it is a double star". This is quite a disparaging remark. But why did the Hipparcos project invent new identifiers?

A. We had (at least) three options:

(a) to introduce a new identifier corresponding to this new catalogue, following general catalogue construction traditions. This had numerous benefits throughout the mission preparation and data analysis phase, but also led to a unique number for each object, a simple classifier which could be referred to uniquely, and a convenient form of connection between the main catalogue, its various annexes, and throughout the documentation;

(b) drop all identifiers, and use only the object's position at some specified epoch and within some specified reference system. This would have been logical (and is an approach used in other surveys). It was considered, but it would probably have been widely criticised for a number of reasons, well-known to those who have reflected on the problem. In any case, this solution would have been equally guilty of renaming well-known stars with yet another designation;

(c) we are left with the approach presumably implicitly recommended in the above remarks by Griffin et al: to use identifiers from the HD Catalogue, for example. However, such identifiers are not available for a large number of objects in the catalogue (they have the additional inconvenience for a printed catalogue of being only tolerably correlated with the object's right ascension). So, presumably, they would need to be supplemented by identifiers from, say, DM. Since these are also not available for all objects (and are subject to numerous errors, including those propagated through the literature) they would presumably, in turn, need to be supplemented by SAO numbers, or by Gliese numbers, or by LHS/LTT numbers; or by whatever identifier you believe constitutes the object's "well-known" designation. In practice, the Input Catalogue allocated 8 fields for these major cross-identifiers. The Hipparcos project felt it inappropriate to allocate the same (or other) fields to reproduce these cross-identifiers, which are unrelated to the Hipparcos mission, and which (again) cannot easily be searched in a printed catalogue. The problem might have been circumvented if we'd not chosen to print the Hipparcos Catalogue (the identifiers could have been built - and frozen - into the machine-readable version) but we believe this would have made the catalogue less easy to access and use.

There is another complicating factor, which is that a Hipparcos entry is not necessarily a (single) star: thus the fact that a HIP entry is recognized as a double star is more efficiently conveyed by the use of a HIP number than it could be by manipulating the HD number(s) of the star(s). And in the words of Daniel Egret of the CDS: "As a data center we are certainly supportive of the strategy chosen by the Hipparcos project, which implied giving a running number to all Hipparcos entries: this is very valuable to help us provide a pointer to Hipparcos data in the databases."

Meanwhile, to get a feeling for how well-known the stars "renamed" by Hipparcos actually are, and to illustrate the problem of cross-identifications, try a little test. Enter the SIMBAD data base using Griffin et al's reference, HR 51565/6. SIMBAD returns "This identifier is not correctly written: HR 51565". In this case the authors were evidently refering to HD 51565/6 - a typo, and there are all-to-many of them in the literature!

Q. Can you explain the standard errors that you give with the catalogue measurements? [Mar98]

A. The Hipparcos & Tycho Catalogues adopt the following formal definition of the "Standard Error", as given in Appendix A (Glossary) of Volume 1:

Standard error: estimated errors are given following the recommendations of the `ISO Standards Handbook 3: Statistical Methods', where the standard error is defined as: `the standard deviation of an estimator; the standard error provides an estimation of the random part of the total estimation error involved in estimating a population parameter from a sample.'

In practice, the characterisation of the random part of the error assumes specific properties of the error distributions, viz. that they are normal (Gaussian) distributed. This is explained in Section 1.5.2 (Error Propagation) of Volume 1:

In Section 1.2.8 a standard model of stellar motion was introduced, and other, more complex models are discussed in Section 2.3. The numerical fitting of such a model to observational data generally results in an estimate both of the parameter vector a and of its covariance matrix C . With n free parameters, these are expressed respectively as matrices of dimension n×1 and n×n.

[From a statistical viewpoint, a and C jointly provide a complete specification of the fitted model and its uncertainties only in the context of linear estimation and normal (Gaussian) error distributions. The estimation problems encountered in the Hipparcos and Tycho data analyses are seldom strictly linear, and sometimes strongly non-linear, and the error distributions are in practice never Gaussian. Perhaps the most dramatic consequence of non-linearity is the possible existence of grid-step errors, which in unfortunate cases may result in positional errors of the order of an arcsec while the formal standard errors remain in the milliarcsec range. Additional statistics provided in the catalogue, such as the rejection rate and goodness-of-fit (Fields H29-30), provide some indication of the sometimes abnormal behaviour of data, but in practice no complete characterisation of the fit is possible. Nevertheless, in the vast majority of cases a and C provide an extremely useful approximation to complex reality.]

The resulting probability levels can be assessed from standard tables of the normal probability integral (see, e.g. tabulations in Ambramowitz & Stegun). Thus, for example, on the assumption that the errors are normally distributed, the probability that the true quantity falls between ± 1 sigma is 0.6827, the probability that it falls between ± 2 sigma is 0.9545, etc.

The caveats noted above should be kept in mind in drawing astrophysical conclusions from these formal probabilities.

The same definitions indeed apply to all other astrometric parameters throughout the Hipparcos and Tycho Catalogues.