arX

iv:a

stro

-ph/

0402

411v

1 1

7 Fe

b 20

04

The Southern Proper Motion Program III.

A Near-Complete Catalog to V=17.5

Terrence M. Girard, Dana I. Dinescu, William F. van AltenaAstronomy Department, Yale University, P.O. Box 208101, New Haven, CT 06520-8101

girard@astro.yale.edu

Imants PlataisThe Johns Hopkins University, Department of Physics and Astronomy, 3400 N. Charles St., Baltimore,

MD 21218

David G. MonetUS Naval Observatory, Flagstaff Station, P.O. Box 1149, Flagstaff, AZ 86002

Carlos E. LopezUniversidad de San Juan and Yale Southern Obs., Av. Benavidez 8175 Oeste, Chimbas, 5413 San Juan,

Argentina

ABSTRACT

We present the third installment of the Yale/San Juan Southern Proper Motion Catalog,SPM3. Absolute proper motions, positions, and photographic B,V photometry are given forroughly 10.7 million objects, primarily stars, down to a magnitude of V=17.5. The Catalog cov-ers an irregular area of 3700 square degrees, between the declinations of -20◦ and -45◦, excludingthe Galactic plane. The proper-motion precision, for well-measured stars, is estimated to be 4.0mas yr−1. Unlike previous releases of the SPM Catalog, the proper motions are on the Inter-national Celestial Reference System by way of Hipparcos Catalog stars, and have an estimatedsystematic uncertainty of 0.4 mas yr−1. The SPM3 Catalog is available via electronic transfer. 1

As an example of the potential of the SPM3 proper motions, we examine the Galactocentricvelocities of a group of metal-poor, main-sequence A stars. The majority of these exhibit thick-disk kinematics, lending support to their interpretation as thick-disk blue stragglers, as opposedto being an accreted component.

Subject headings: astrometry – catalogs

1. Introduction

The Yale/San Juan Southern Proper Motion(SPM) Program was originally intended as asouthern-sky complement to the Lick NorthernProper Motion (NPM) Program (Klemola etal. 1987). Photographic B,V photometry, posi-tions and absolute proper motions would be de-termined for a significant fraction of the southern

sky, as outlined by van Altena et al. (1990, 1994).The photographic plates, upon which the SPM isbased, extend to approximately V=18. Previousversions of the SPM Catalog were derived fromPDS microdensitometer measures of these plates.Limited by the throughput of the PDS system, in-clusion of all measureable images on the plates wasimpossible. Instead, an input list was compiled,consisting of stars of special interest gleaned from

1

the literature, all measureable galaxies (needed forthe absolute proper-motion reference frame), anda randomly selected sample of stars intended forkinematic study.

Such an input list was used to construct thefirst SPM Catalog which covered an area of 720square degrees at the South Galactic Pole (SGP)and included roughly 50,000 stars and 9,000 galax-ies, (Platais et al. 1998, hereafter Paper II). Thesecond SPM Catalog extended this area to a 3700-square-degree band in declination from approxi-mately -20◦ to -45◦, but excluding the Galacticplane fields. This second catalog, the SPM2 2,also included a re-reduction of the measures thatcomprised the SPM1 Catalog. The SPM2 con-tained 287,000 stars and 35,000 extragalactic ob-jects. This represents less than 3% of the totalnumber of measureable stars on these plates.

The present catalog, SPM3, is based on thesame plates and covers the same area as the SPM2.However, the plates have now been rescannedusing the Precision Measuring Machine (PMM)of the US Naval Observatory, Flagstaff Station,(Monet et al. 2003 includes a description). Thisallows all identifiable images to be measured, al-beit to a slightly reduced precision relative to thatattainable with the Yale PDS. The result, is a cat-alog of absolute proper motions, positions, andB,V photometry that is approximately completeto V=17.5. (We estimate the incompleteness dueto confusion caused by overlapping images andunidentified high proper-motion stars to be 5%.)The SPM3 contains a total of 10.7 million objects,over the same 3700 square degrees covered by theSPM2.

As with earlier versions of the SPM Catalog,an explicit correction has been made for magni-tude equation – the magnitude-dependent system-atic shift in image position that afflicts all pho-tographic material to some extent. Among thevarious photographically based proper-motion cat-alogs currently available, the SPM is unique inthis important regard. Magnitude equation, andits correction, is discussed in detail by Girard etal. 1998, (hereafter, Paper I). Whereas earlier ver-sions of the SPM Catalog used external galaxies toset the zero-point of the absolute proper-motionframe, the SPM3 is on the proper-motion system

2described in detail at http://www.astro.yale.edu/astrom/

of the International Celestial Reference System(ICRS) via the Hipparcos Catalog (ESA 1997).

In the following section, the plate material andplate measurement are described. In Section 3, theastrometric and photometric reduction proceduresand catalog construction are detailed. In Section4, the properties of the catalog are examined bymeans of internal and external comparisons. Thevarious incarnations of the SPM Catalog are de-scribed in Section 5, to assist the potential user inselecting the appropriate version. Section 6 illus-trates a practical application of the SPM3 Catalog– using the kinematic data to better determine thenature of a group of metal-poor, main-sequence Astars, first pointed out by Rodgers (1971). Thefinal section summarizes the current status andfuture direction of the SPM program.

2. Plate Material and Measurement

Observations were carried out using the 51-cmdouble astrograph of the Cesco Observatory in ElLeoncito, Argentina. The double astrograph is es-sentially two telescopes sharing a common tube.One of the objective lenses is optimized for thevisual passband, the other for blue. Originally,the SPM program was to extend from the southcelestial pole to the equator, providing substan-tial overlap with the NPM program that extendsfrom the north down to roughly -23◦ declination.Due to limited resources, the first epoch of theSPM program was halted after coverage from thesouthern celestial pole to about -20◦ was com-pleted, roughly 76 percent of the southern hemi-sphere. The first-epoch plates were taken in thelate 1960’s to early 1970’s. The second-epoch sur-vey was begun in the late 1980’s, but was onlyabout 35 percent completed when Kodak discon-tinued production of the 103-a emulsions. Most ofthe SPM fields for which second-epoch plates existhave been included in the SPM3, with the excep-tion of the low Galactic-latitude fields. The SPM3Catalog is based on 156 such SPM fields. SPM skycoverage is illustrated in Fig. 1. The mean epochof the first-epoch plates is 1968.0 and that of thesecond-epoch is 1991.4.

With a scale of 55.′′1 mm−1, the 17-inch photo-graphic plates used cover an area of 6.3◦ × 6.3◦.They are taken on 5◦ centers in declination, withvarying offsets in right ascension, but never ex-

2

ceeding 5◦. In most cases, the blue and visualplates for a specific field were exposed simultane-ously. Each plate contained two exposures, oneof 2 hours duration, and an offset exposure of 2minutes duration. Each of the exposures was cen-tered on the meridian. Also, an objective grat-ing, with grating constant ∆m=3.85, was used toproduce diffraction-image pairs that bracket thebrighter central-order images. These grating im-ages not only extend the useful dynamic range ofthe plates but also provide a means for calibrationand correction of each plate’s magnitude equation.

The blue and visual pairs of first and second-epoch plates were scanned with the Precision Mea-suring Machine (PMM) of the USNO FlagstaffStation. We present here only a brief descriptionof the PMM hardware and software. For a morecomplete description, see Monet et al. (2003) andreferences therein.

The PMM uses a 2-dimensional CCD detec-tor (1312×1033 useful pixels) to snap a series of“footprint” images of the photographic plate. Theplaten, holding the plate, is repositioned preciselybefore each exposure in a manner such that the en-tire 17-inch plate is scanned by an array of 26×34such footprints. There is a small amount of over-lap between adjacent footprints. The effectivepixel size is 0.′′75.

Each footprint, consisting of an array of trans-mission values, is processed in real time by thePMM software pipeline. A “blob”-detecting rou-tine identifies groups of pixels above the locally-determined sky background. A 5-parameter func-tional fit is then performed, by least squares, to thetransmission values, T (x, y), minus background,BT . The circularly symmetric function used inthe PMM pipeline is the following,

T (x, y)−BT = A[eα((x−x0)2+(y−y0)

2−r2

0) + 1]−1

(1)where A is the image’s amplitude, x0 and y0 areits position, and r0 and α represent a saturationradius and a measure of the wings’ extent, respec-tively. The values of x and y within the footprintare simply the CCD pixel coordinates multipliedby the nominal pixel scale.

The position of the CCD at the time the foot-print is taken, is measured accurately by a laserinterferometer. This is combined with the derivedimage center, (x0, y0), to yield the plate coordi-

nates of each image detected and centered. Atthe time the SPM plates were scanned, the PMMpipeline made a scalar correction to the nominalpixel scale in each footprint. We found this wasinadequate and instead determined a general lin-ear correction, (with relative scale, rotation, andskew terms), per footprint by minimizing the dif-ferences in position for multiply measured starsin the overlap regions of the footprints. (Subse-quently the PMM pipeline has also been modifiedto include a field-dependent “refocus” correctionbased on the overlap images.)

Finally, the PMM software calculates addi-tional image parameters. These include a mag-nitude estimate based on the integrated flux ofthe pixels used to perform the functional fit. It isthis instrumental magnitude estimate which willbe used in our photometric reduction.

3. SPM Catalog Construction

The goal of the SPM3 is to be as complete aspossible to a limiting magnitude determined bythe plate material. The most obvious approach isto construct the catalog from the complete list ofimages detected by the PMM software, setting acutoff at some specified signal-to-noise threshholdlevel. Unfortunately, the presence of the gratingimages, (in both long and short exposures), as wellas a large number of spurious detections and multi-ple detections, make it difficult to implement sucha straightforward approach.

Instead, we have chosen to first construct aninput catalog from existing, external catalogs;namely the Hipparcos Catalog (ESA 1997), theTycho-2 Catalog (Hog et al. 2000), the UCAC0.9 (see Zacharias et al. 2000 for a description ofthe UCAC1), and the Two Micron All Sky Survey(2MASS) point-source and extended-source cata-logs. The external catalogs are merged, in the or-der listed above, to provide an input catalog thatis as complete as possible.

Of course, these input catalogs contain substan-tial overlap, with almost all Hipparcos stars be-ing in the Tycho-2 catalog, and the majority ofthese, as well as almost all UCAC stars, being con-tained in the 2MASS catalogs. For the purposesof preparing our master input catalog, duplicateswere identified by simple positional coincidence us-ing a matching tolerance of 1.1 arcseconds.

3

With the input catalog in hand, the next stepis to identify each PMM-detected image with animage system (e.g. long-exposure, central order)and a specific source from the input catalog. Thisis accomplished by forming vector differences be-tween the x,y coordinates of the brightest PMM-measured images and the predicted plate coordi-nates of Tycho-2 stars, knowing the plate scale,orientation, and approximate plate-center tangentpoint. The discrete clusters of points in thisvector-difference space make it possible to identifythe various exposure/grating systems, and deter-mine their relative separations. Based on thesepreliminary image-system identifications and thepredicted plate coordinates of the Tycho-2 starsin the input catalog, approximate transformationsare found relating each of the image systems tocelestial coordinates. The transformations areapplied to each image system and an improvedmatching with the Tycho-2 stars is made by posi-tion, and from these matches, a new set of image-system transformations is derived. The form ofthe final transformation equations is a general cu-bic polynomial.

Applying the transformations to all images de-tected on the plate, a spatial matching is per-formed with the full input catalog. A rather largematching tolerance of 100 µm (5.5 arcsec) is used.(A cleaning of improperly matched objects is madelater.) All measured images, through third-orderin both long and short exposure, for each inputcatalog star are extracted and saved, along withtheir input-catalog identification. Once the mea-surements have been extracted for all four platesin an SPM field, (B and V plate pairs at first andsecond epoch), the photometric and astrometricprocessing is performed on a field by field basis.

3.1. Photometric Reduction

The PMM software provides an instrumentalmagnitude index, p, for each image detected andcentered. The calibration of these photometric in-dices to B, V estimates is a four-step process.

First, the indices are made nominally linearwith true magnitude by use of the grating con-stant, which is assumed to be field and magnitudeindependent. A polynomial, of up to fifth-order,plus magnitude-field terms is adopted for the form

of the “linear” magnitude, m.

m = a1p+ a2p2 + a3p

3 + a4p4 + a5p

5

+a6px+ a7 (2)

The coefficients, ak, are determined by leastsquares using the condition that the difference be-tween the first-order and central-order magnitudeestimates for all stars, for which both can be mea-sured, should be equal to the grating constant. Avalue of m(n=1)−m(n=0) = 3.85 is adopted for thegrating constant. A separate set of ak are deter-mined in the long and short-exposure systems, andthen applied to all images in that system. Notethat had a constant term been included in Equa-tion (2), it would be indeterminable, as would purefield terms. Also, any deviation in the actual grat-ing constant from the nominal value of 3.85 willresult in a non-unity scale term in the linearizedmagnitudes. All of these terms, which cannot bedetermined at this stage, are included in the thirdstep of the photometric reduction process.

The second step in the photometric reductionplaces the short and long- exposure photometricindices onto a common system. Due to the differ-ences in the way in which the stellar-image profilesvary across the plates in these two systems, thistransformation allows for the inclusion of up to cu-bic field terms, when they are deemed significant.

mlong −mshort = b1 + b2x+ b3y + b4x2 + b5xy

+b6y2 + b7x

3 + b8x2y + b9xy

2 + b10y3 (3)

In actuality, the constant term in the above ex-pression is redundant, but is included for compu-tational convenience.

After application of the above transformationto the short-exposure images, the photometric in-dices of all measured grating orders in both expo-sures, m, will be on a common system, to within aconstant, i.e., not a function of magnitude. Whatremains is to determine the constants, for the vari-ous grating and exposure systems, i.e., the magni-tude difference between the central-order and var-ious higher diffraction orders in both the long andshort exposures, and any field dependence theseconstants might have. This is done in the thirdstep of the photometry calibration process, usingcalibrating reference stars of known magnitudes.

mref = c1,k+c2x+c3y+c4x2+c5xy+c6y

2+dmk

(4)

4

where k =1 to 8 indicates the grating-order/exposuresystem, central order through third order in boththe long and short exposure. The referencestar magnitudes were originally taken from threesources – the Guide Star Photometric Catalog(Lasker et al. 1988); a B, V CCD sequence takenwith the CTIO 0.9-m telescope at the center ofeach of our SPM fields, specifically taken for theconstruction of earlier versions of the SPM Cat-alog, (see Paper II); and Tycho-2 Catalog B, V

values, in order to provide the necessary leverageon the field terms. After some experimentation,we found that the faint end of the calibration couldbe better determined by also including the (PDS-measured) SPM2 Catalog stars as photometriccalibrators, providing thousands of reference starsper plate. Subsequent to this calibration step,multiple magnitude estimates of the same star,from different exposure/diffraction systems, arecombined by weighted average using a schemesimilar to that detailed in Paper II. This gives asingle B and V estimate per star per epoch.

The first and second-epoch magnitude esti-mates are averaged to yield individual B and V

estimates per star. Uncertainties are derived fromthe scatter in the magnitude estimates from thevarious exposure/diffraction systems and from thetwo different epochs.

Even with the additional SPM2 photometricreference stars, there remained systematic differ-ences as a function of magnitude when comparedto the SPM2 photometry. The bottom panel inFig. 2 shows a comparison between SPM2 andpreliminary SPM3 B-magnitude estimates, for asample SPM field. The differences in V are notshown but are similar to those in B. The largedeviations in the magnitude calibrations are un-doubtedly rooted in the raw photometric indices.There is a significant difference in the forms of theimage-fitting functions used to reduce the PMMmeasures of the SPM3 and that of the PDS mea-sures of the SPM1 and SPM2. As our photometriccalibration pipeline was originally designed for thePDS measures, it is likely the simple applicationof the same routines to the PMM measures thataccounts for the severe magnitude trends in thederived photometry and the SPM2 data are to beconsidered more reliable.

Thus, as a last step in the photometric reduc-tion of the SPM3 Catalog, a final correction was

applied to each star’s B and V estimate. Thiscorrection was a moving median, as a function ofmagnitude, of the differences SPM2 photometryminus SPM3 preliminary photometry, on a fieldby field basis. For each star in the SPM3 inputlist, the median difference of the 50 stars nearestin magnitude to the target star and in commonwith the SPM2 was used as a correction to theSPM3 magnitude. The bottom panel in Fig. 3shows the differences between SPM2 and SPM3photometry after the moving median correctionshave been applied, for the same sample field shownin Fig. 2. The scatter on the bright end is disap-pointingly large, but the systematic trends havebeen effectively removed. (The diagonal bandingin the bottom panel of Fig. 3 is an artifact causedby the use of the SPM2 data both to adjust thephotometry and to assess it, as a function of thecorrected photometry itself.)

3.2. Astrometric Reduction

The astrometric reduction procedure of theSPM3 differs from that of previous versions ofthe SPM Catalog, but effectively achieves simi-lar results by making use of the SPM2 data tocalibrate the final astrometry. In the past, rela-tive proper motions were derived by using a largenumber of faint, anonymous stars to transformmeasures from a first-epoch plate into the sys-tem of the same passband second-epoch plate –a differential method. The correction to absoluteproper motions was then a matter of using thegalaxies’ relative proper motions to determine theconstant offsets, field by field.

For logistical reasons, preliminary SPM3 propermotions are derived directly from differences inequatorial coordinates determined separately atSPM first and second epoch. Tycho-2 stars areused to determine the polynomial plate-model co-efficients. In theory, this will yield proper mo-tions on the system of the ICRS. However, thenumber of high-quality Tycho-2 stars per fieldis insufficient to adequately determine the plate-model coefficients, producing systematic errors inthe proper motions, (i.e. modeling errors), as afunction of position. These are removed by apply-ing field-dependent adjustments calculated fromthe thousands of SPM2 stars per field, effectivelytransforming the proper motions to the (differen-tially determined) system of the SPM2.

5

As pointed out in Table 3 of Paper II, therewere small but significant differences between theabsolute proper motions of SPM1 and the Hippar-cos Catalog. These we believe can be attributed toresidual magnitude equation in the “motions” ofthe galaxies used to set the proper-motion zeropoint of the SPM1 and SPM2 Catalogs. Thegalaxy image profiles, of course, differ from thestellar profiles upon which the magnitude equationcorrections are based. An attempt was made to ac-count for the expected difference in the form of thegalaxies’ magnitude equation correction, (see Pa-per I), but some residual error no doubt remained.For this reason, we have chosen to use Hippar-cos Catalog stars instead to set the zero point ofthe absolute proper motions in the SPM3. Thefollowing is a more detailed outline of the SPM3Catalog’s astrometry reduction pipeline.

The first step in the astrometry reduction isa pre-correction for differential refraction, acrossthe 6.3 ◦ field of view. Differential color refractioncorrection is not deemed necessary, as the platesare exposed centered on the meridian.

The next step is to solve for the magnitudeequation correction, and simultaneously transformthe grating images onto the system of the central-order images. The procedure used is identicalto that described in Paper I, using central andfirst-order image pairs to derive the coefficients ofa polynomial magnitude-equation correction, al-though in this instance, we have confined the func-tion to quadratic field terms plus a quadratic mag-nitude term.

∆X = F (m,X, Y ), ∆Y = G(m,X, Y ) (5)

where

F = f1m+ f2mX + f3mY

+ f4mX2 + f5mXY + f6mY 2 + f7m2,

G = g1m+ g2mX + g3mY

+ g4mX2 + g5mXY + g6mY 2 + g7m2

(6)

In the above, linearized instrumental magni-tudes are used, i.e., after application of the pho-tometric coefficients derived in Equations 2 and3. The coefficients, fk and gk, are found by leastsquares minimization of the differences in positionbetween the central order and the average of thefirst-order pair of images, for all stars for which

these three images can be measured.

X(n=1) −X(n=0) =F (m(n=1), X, Y )− F (m(n=0), X, Y )

Y(n=1) − Y(n=0) =G(m(n=1), X, Y )−G(m(n=0), X, Y )

(7)

Separate sets of coefficients are derived for thelong and short-exposure systems, with typicallythousands of stars used in the long-exposure so-lution and hundreds in the short. These are thenapplied to all images in each exposure system, cen-tral through third diffraction order.

We note that the same magnitude-equation cor-rection has been applied regardless of whether theimages are of stars or galaxies, while the correc-tion is derived entirely from stellar images. Forthis reason, the proper motions of those objectsidentified as extended sources in the SPM3 shouldbe used with caution.

After application of the magnitude-equationcorrection and averaging the positions of gratingpairs, all images are on one of two systems; thelong-exposure system, or the short-exposure sys-tem. These two are combined using yet anotherpolynomial transformation, this time a general cu-bic model. The number of stars used to deter-mine this “bridge” solution was on average 6500per field.

With all X,Y measures now on a single sys-tem, the transformation to celestial coordinatescan be calculated in a straightforward manner.Tycho-2 stars, trimmed somewhat by positionaland proper-motion error, are used as referencestars. These are updated to the epoch of eachSPM plate and then transformed to standard coor-dinates using the nominal field centers. As demon-strated by Platais et al. (1995), cubic field termsare necessary to properly describe the SPM plates.Thus, a general third-order polynomial transfor-mation between plate coordinates and standardcoordinates was assumed, i.e., a 10-term solutionin each coordinate. The number of Tycho-2 refer-ence stars varied from 170 to 1100, with an averageof 400 per field. The standard error of unit weightper coordinate ranged from 65 to 200 mas, with amode of 80 mas.

For stars with multiple grating and/or exposure-system measures, a single position is determinedfrom the weighted average of the individual de-

6

rived positions. There is simultaneously a trim-ming of discordant position estimates. Theweighting and trimming procedures, as well asthe procedure for estimating internal uncertain-ties, are similar to those used in previous SPMreductions, as described in Paper II. This yieldspositions, in general, from both the blue and vi-sual plates at first and second epochs. Withineach epoch, the blue- and visual-derived positionsare averaged, weighted by the number of imagesper plate for each star.

In theory, having positions on the system of theICRS at two epochs, one need only take differ-ences and divide by the epoch difference to de-rive absolute proper motions. This we do, but theresulting proper motions are not considered final.In practice, the high-order polynomial transforma-tion and the often insufficient number of Tycho-2reference stars, can result in significant system-atic effects in the derived positions and proper mo-tions.

The top two panels of Fig. 2 show differencesbetween the so-derived preliminary positions andproper motions compared to those of the SPM2Catalog, as a function of plate coordinate, for asample SPM field. We show only the right as-cension components of the position and proper-motion differences as a function of x coordinate,but the trends in the declination component andas a function of y coordinate are similar in ap-pearance. The position-dependent effects seen arenot unexpected. The proper motions of the SPM2were derived in a differential manner, using a largenumber of anonymous faint stars to determinethe difference between the first and second-epochplates’ models. Thus, they are better determinedthan these preliminary SPM3 proper motions. Asfor the trends in positional differences, these arelikely due to statistical uncertainty in the polyno-mial coefficients derived from different sets of ref-erence stars; Tycho stars in the case of SPM2, andTycho-2 stars in the case of SPM3. In both cases,the number of reference stars is only marginallysufficient to determine the polynomial plate mod-els. Fortunately, the large number of SPM2 starsin the SPM3 sample allows us to calibrate the pre-liminary SPM3 data onto the system of the SPM2.

The very differences shown in Fig. 2 are used toperform this calibration. For both the SPM3 posi-tions and proper motions, separately, a correction

is determined based on the median difference withthe corresponding SPM2 data. The 50 SPM2 starsnearest to each target SPM3 star (spatially, by 2-dimensional separation) are used in determiningthe median correction vectors. The medians arecalculated and applied star by star, one SPM fieldat a time, i.e., the corrections vary from field tofield. The top two panels of Fig. 3 show the differ-ences with SPM2 after the corrections have beenapplied to the SPM3 astrometry.

At this stage, the SPM3 positions and propermotions are on the system of the SPM2. Thepositions in the SPM2 itself, however, are knownto suffer from a systematic field error when com-pared to Tycho-2 positions. This was attributedto the inadequacy of the third-order polynomialused to transform the SPM plate measures intocelestial coordinates, during construction of theSPM2 Catalog. The SPM3 positions, being on thesystem of the SPM2, are expected to suffer simi-larly and, thus, we shall make an explicit correc-tion for this systematic field pattern. The SPM2proper motions, being differential, should not suf-fer from any such effects, provided the telescopeand plates remained unchanged between the twoepochs of observation. Prompted by the refereeof an earlier draft of this paper that describedan earlier version of the SPM3 Catalog, we haveinvestigated the proper motions of the SPM3 forany such field pattern as well. We were surprisedto find that one exists. It implies that a system-atic change has occurred in either the telescope orthe stored plates, causing a difference in the formof the transformation from sky coordinates intoplate coordinates between first and second epoch.Whatever its cause, the systematic effect in propermotions must be corrected for as well.

The corrections, made separately for the po-sitions and proper motions, are based on resid-ual “masks”. In the case of the proper motions,residuals are formed between the Tycho-2 valuesand the SPM3 proper motions. The differencesare binned as a function of position within eachSPM field, (i.e., by plate position), onto a 21×21grid. The mean proper-motion differences, uponstacking the 156 SPM fields, are shown in Fig. 4a.Bi-linear interpolation within the grid is used tocalculate the correction to be applied to each ob-ject in the SPM3 Catalog.

In the case of the positional correction mask,

7

residuals are formed using the UCAC 0.9 Cata-log and the SPM3. The much higher density ofUCAC over Tycho-2 provides a better determina-tion of the position vector residuals. The meanposition differences are shown in Fig. 4b. In cal-culating the residuals, the SPM3 proper motionsare first mask-corrected and then used to bringthe SPM3 positions to the UCAC epoch. As withthe proper-motion correction, bi-linear interpola-tion is used to calculate individual corrections tobe applied. And as with the proper-motion mask,the mean correction across the entire field is forcedto zero, by means of a constant offset, thus keepingthe field-by-field mean position and mean propermotion on the same system as the SPM2. The po-sitions in the SPM2 are on the 1991.25 system ofthe original Tycho Catalog. Thus, the SPM3 posi-tions will be on this same system, which in effect,is the same as that of Tycho-2.

The proper-motion system of SPM2 is indepen-dent of Hipparcos and Tycho-2. As stated earlier,and detailed in Paper II, the proper-motion zeropoint of each SPM2 field was determined from themean relative motion of the galaxies in that field.The magnitude equation affecting the galaxy im-ages is expected to be different in form from thataffecting the stellar images, upon which the cor-rections are based. This was accounted for, tofirst-order, in earlier SPM reductions by solvingfor an effective magnitude offset between the stel-lar and galactic magnitude equation corrections,(see Paper I).

Realizing this simple compensation may beinadequate, especially for the PMM-measuredgalaxies, we have decided to calculate the abso-lute proper-motion zero point of the SPM3 fromHipparcos Catalog stars, and their Hipparcos-measured absolute proper motions. For eachSPM3 field, the correction to absolute proper mo-tion is the median of the differences in proper mo-tion between the Hipparcos values and the SPM3measures, after being calibrated to the SPM2 sys-tem. We note that the proper-motion systemof Hipparcos is less likely to be affected by un-corrected magnitude equation compared to thatof Tycho-2, which relies on ground-based pho-tographic material. For this reason, we chooseto use Hipparcos to determine the absolute zero-point corrections of the SPM3 proper motions,field by field.

Fig. 5 shows the distribution of the correctionsfor the 156 fields that comprise the SPM3. Eachpoint represents the correction applied to a partic-ular field. The mean correction is effectively zeroin µα∗ = µαcosδ, but in µδ it is +0.55 ± 0.12mas yr−1. This is similar to what was found ina comparison of SPM1 proper motions with thoseof Hipparcos, described in Paper II. The numberof Hipparcos stars per field varies from 63 to 147,with an average of 102. As will be shown in Sec-tion 4, the precision of the SPM3 proper motionsat these magnitudes is roughly 4 mas yr−1. Thus,the accuracy to which the final SPM3 proper mo-tions are on the absolute system of Hipparcos isexpected to be 0.4 mas yr−1.

3.3. Compilation

At the completion of the procedures describedabove, calibrated B, V photometry, celestial co-ordinates, and absolute proper motions will havebeen determined for all objects in the input list,for each of the 156 SPM3 fields. What remains isto combine the data from all fields. First, though,a culling is made of improperly identified objects.

The SPM3’s PMM-measured detections wereidentified by simple positional coincidence withthe four input catalogs, dominated by the 2MASSpoint source and extended source catalogs. Withno previously determined proper motions for thevast majority of the target objects, a relativelylarge positional coincidence tolerance of 5.5 arcsecwas used to match target objects with PMM de-tections. This led to false identifications in somecases, in one or both epochs, due to the large num-ber of spurious detections in the raw PMM mea-sures.

Therefore, a final check of positional coinci-dence is made, this time at epoch 2000. All SPM3objects are updated to this epoch, using the SPM3proper motions, and compared with their inputcatalog positions. A matching tolerance of 2 arcsecis enforced. Roughly 8% of the tentatively iden-tified objects were determined to be spurious andeliminated.

The final step is the merging of the 156 fields,averaging the measures of objects in the overlaparea of adjacent fields. Again, a 2-arcsec posi-tional matching tolerance is used for this purpose.Averages are weighted, based on the estimated un-

8

certainties in the quantities being averaged.

4. Catalog Properties

The SPM3 Catalog includes 10,764,865 objects.It contains positions and absolute proper motions,on the system of the ICRS, and photographic B, V

photometry, and estimated uncertainties for eachof these quantities. The positions are given atepoch 1991.25, the mean epoch of the Hippar-cos and Tycho Catalogs and very nearly the meanepoch of our second-epoch plate material. Thepositions and proper motions have been used tomake a cross-referencing to other selected catalogs,including the original input source catalogs, bypositional coincidence. This provides both com-plementary data, e.g. Hipparcos parallaxes, andcomparable data, e.g. Tycho-2 proper motions.Table 1 summarizes the number of objects in theSPM3 which have been cross-identified in othercatalogs. The cross-identification data are avail-able along with the SPM3 Catalog via the web, athttp://www.astro.yale.edu/astrom/.

The uncertainties in the SPM3 Catalog are de-rived from multiple measures of the same object,in different exposure systems and/or diffractionorders and/or passbands and/or fields, for starsin the overlap region of adjacent SPM fields. Forthose objects with only one image and one mea-surement per plate, a value is assigned based onthe mean uncertainty of other (multiply measured)images with similar signal-to-noise. Fig. 6 illus-trates the Catalog uncertainties as a function ofmagnitude. Only uncertainties along right ascen-sion are shown, to save space. The uncertaintiesalong declination exhibit similar trends. The un-certainty estimates are quantized, particularly thephotometric estimates. For the purpose of plot-ting, the uncertainties shown in Fig. 6 have beensmoothed slightly. The median uncertainty, as afunction of magnitude, is shown for the positionsand proper motions. The photometric uncertain-ties are severely quantized due to the adoption ofa baselevel uncertainty of 0.25 mag early in thereduction procedure, during the magnitude lin-earization. Consequently, displaying the medianof the photometric uncertainties is not instructive.Instead, a comparison is made with the CCD pho-tometry of over 7000 stars used in the calibrationof the photographic photometry. The dotted curve

in the bottom panel of Fig. 6 shows the disper-sion of the differences with this CCD photometry.Also, an external error estimate in proper motionis made at the bright and faint ends of the Catalogdistribution.

Based on approximately 11,000 Hipparcosstars, the internal proper-motion uncertaintiesunderestimate the external values by about 30percent. At the faint end, the dispersion in propermotion of roughly 300 quasars, (originally identi-fied in the SPM2), is 12 mas yr−1, significantlylarger than the internal error estimate. No exter-nal comparison is made of the SPM3 positions,but, as with the SPM2 Catalog, the internal un-certainty estimates most likely also underestimatethe external values by about 30 percent. Thus,we estimate the positional precision to be 40 masand the proper-motion precision to be 4 mas yr−1,for well-measured stars in the SPM3. With ap-proximately 100 Hipparcos stars per field availableto set the proper-motion zero point, we estimatethe accuracy of the SPM3 absolute proper-motionsystem to be 0.4 mas yr−1. The photometric pre-cision is ∼0.15 magnitudes, for the best-measuredstars.

The input catalog from which the list of SPM3objects is derived is expected to be nearly com-plete to the limiting magnitude of the 2MASSPoint Source Catalog. Of course, near this limit,the actual completeness of the SPM3 will be lower.Issues of image confusion and PMM detectionidiosyncrasies affect the catalog completeness atall magnitudes, but will be most severe for thefaintest stars. Fig. 7 indicates the magnitude dis-tribution of the final SPM3 Catalog. For compari-son, also shown are the distributions of other cata-logs over the same area of sky, including the PDS-based SPM2 Catalog, which was not designed tobe complete to any magnitude. The falloff of theSPM3 distribution beyond V=17.5 is remarkablyabrupt, given that it must reflect both the vari-ation in plate depth among the 156 fields and abroadening by the magnitude uncertainties.

It is not a trivial task to quantify the complete-ness of the SPM3 as a function of magnitude with-out an external catalog that is known to be 100%complete to a deeper limit and in a comparablepassband. However, we can make an approximateestimate of the relative completeness between theSPM3 and the 2MASS point source catalog. In

9

order to make a direct comparison, as a functionof SPM3 magnitude, we require an approximatetransformation from 2MASS J,H,K magnitudesto V . We use a relation, that we have determinedempirically,

V ≈ J + 2.79(J −K) (8)

and find to be an adequate approximation for alarge range of stellar types. Applying this rela-tion, we calculate and display in Fig. 8 the V -magnitude distribution of the 2MASS point sourcecatalog and compare it to that of the SPM3, in asample SPM field. In the range 8 < V < 17.5, theSPM3 distribution is similar to that of 2MASS,to within the statistical noise. The relative com-pleteness drops from 1.03 at V=17.5 to 0.56 atV=18.0. Thus, the completeness magnitude limitof the SPM3 Catalog is taken to be V=17.5.

Even for stars brighter than this magnitude,image confusion and blending cause a low levelof incompleteness. At the very bright end of themagnitude distribution, this incompleteness canbe estimated using the Hipparcos Catalog. AllHipparcos stars were included in the input targetlist and yet the final number of Hipparcos starsin the SPM3 Catalog, 10,900, is 8% below the ex-pected number of ∼ 11,790. By comparison, theSPM2 Catalog, based on PDS measurements anda far less automated reduction pipeline, contains10,957 Hipparcos stars. A large fraction of thesemissing Hipparcos stars are brighter than V=5,and not measureable on the SPM plates.

Perhaps the best estimate of the fraction ofstars lost because of the more automated SPM3reduction pipeline, and other factors, can be madeusing the SPM2 as a comparison. Assuming a con-servative positional matching tolerance of 2 arcsec,we find 6% of SPM2 stars and galaxies are absentfrom SPM3. Assuming a more liberal 3-arcsec tol-erance, the percentage of absentees falls to 4%.Of these missing stars and galaxies, there is lit-tle dependence on magnitude. However, a signifi-cant fraction of the absent stars have large propermotions, based on the SPM2 measures. For in-stance, of the 14,193 absentees (based on the 3-arcsec matching), 1939 are stars with total propermotions in excess of 180 mas yr−1. This bench-mark value of 180 mas yr−1 is relevant given the∼30-year mean difference between the first-epochSPM plates and the 2MASS Catalog, (the dom-

inant input catalog), and the matching toleranceof 5.5 arcsec. Stars with proper motions in ex-cess of this value, and not in the Hipparcos orTycho-2 Catalogs, would not have been properlyidentified and, thus, are absent from the SPM3.An effort was made to include these high proper-motion stars in the construction of the SPM2, butfor the sake of expediency, they were not specifi-cally included in the SPM3. Thus, we caution thepotential user of the SPM3 Catalog as to the se-vere incompleteness of faint, high-proper motionstars.

5. Catalog Versions

With several versions and subversions of theSPM Catalog in existence, it is prudent to describethese briefly, for the sake of clarity. The originalSPM 1.0 is based on PDS measures of 30 SPMfields around the South Galactic Pole. It coversroughly 720 square degrees and contains 58,880objects. The input catalog for the SPM 1.0 is com-prised of objects of interest gleaned from the lit-erature, utility objects (i.e., stars necessary to thereduction procedure), visually confirmed galaxiesfor proper-motion reference, and a large number ofrandomly-selected anonymous stars intended forkinematic study. The SPM 1.0 is not designed tobe complete over any magnitude range.

The SPM 2.0 is based on a re-reduction of themeasures used to construct the SPM 1.0, in ad-dition to PDS measures of other mid-declinationSPM fields for which first and second-epoch platematerial exists. It includes measures from a totalof 156 SPM fields, covering an area of approxi-mately 3700 square degrees. The fields are fromthe -25, -30, -35, and -40 degree declination bands,excluding fields near the Galactic Plane. The in-put catalog for this version of the SPM Catalogwas constructed in a similar fashion to that of theSPM 1.0. Thus, the SPM 2.0 is also not explic-itly designed to be complete over any magnituderange. The SPM 2.0 contains 321,608 objects.

The reduction procedure for the SPM 2.0 in-corporates one significant improvement over thatof the SPM 1.0, that being the handling of themagnitude-equation correction for galaxies. As anaid to those who had made kinematic studies usingthe SPM 1.0, the SPM 1.1 was created by extract-ing all objects from the SPM 2.0 that fell within

10

the boundary of sky defined by the SPM 1.0. A netgain of 8 objects resulted, yielding 58,887 entriesin the SPM 1.1. The SPM 1.1 is a subset of theSPM 2.0. The identification numbers of objects inthe SPM 1.0 and SPM 1.1 are not the same.

The position system of the SPM 1.0, 1.1 and2.0 is that of the original Tycho Catalog, at epoch1991.25. The absolute proper-motion system ofthese versions is based on PDS measures of galax-ies, on a field by field basis. The SPM 1.0 is de-scribed in Paper II, (Platais et al. 1998). Detailsconcerning the SPM 1.1 and SPM 2.0 can be foundat http://www.astro.yale.edu/astrom/

The SPM 3 is constructed from PMM measuresof the same 156 SPM fields which comprised theSPM 2.0 Catalog. Actually, PMM measures weremade of all existing first and second-epoch SPMplates. As part of a project to construct an in-put catalog for the FAME mission, a reductionpipeline for these PMM data was created at theUS Naval Observatory. This pipeline transformedthe PMM measures into celestial coordinates atplate epoch, and incorporated many of the fea-tures of the reduction procedure of the earlier SPMCatalogs. The input catalog for this pipeline wasmuch more complete, though. It consisted of theunion of the Hipparcos, Tycho-2, and UCAC 0.9Catalogs. Later, in order to reach the SPM platelimit, a magnitude-trimmed version of the USNO-A2 Catalog was added to the list of input sourcecatalogs.

Having this reduction pipeline in place, it wasrealized that it could be used to create a new ver-sion of the SPM Catalog, one nearly complete tothe SPM plate limit. Thus, the reduction pipelinewas modified slightly to allow it to be appliedto the PMM measures of both first and second-epoch SPM plates. (The original US Naval Ob-servatory project made use of only the first-epochSPMmaterial.) With first and second-epoch celes-tial coordinates, preliminary proper motions couldthen be derived, and then transformed to a better-calibrated system, as described in detail elsewherein this paper.

The SPM 3.0 was the first version to be basedon PMM measures in this way. It was made avali-able for only a short time, and directly distributedto only a handful of colleagues. Upon the release ofthe full-sky 2MASS data, it was decided to replacethe USNO-A2 as an input source catalog with the

2MASS Point Source Catalog (which does includethe 2MASS extended sources). This resulted inthe SPM 3.1 Catalog, with improved completenessproperties, along with the added bonus of cross-identification to the 2MASS photometry. This ver-sion of the SPM Catalog was also available only fora brief period and once again directly distributedto only a handful of researchers.

It was the comments of the referee on an earlierversion of this paper which led us to discover thesystematic pattern in the stacked proper-motionresiduals of the SPM 3.1. As described in Section3, this systematic error now has been corrected,resulting in the current version of the catalog pre-sented here, SPM 3.2. The SPM 3.2 supercedesboth the SPM 3.0 and SPM 3.1. It differs fromthe previous major version of the SPM Catalog,the SPM 2.0, in two significant ways: The useof PMM measures, as opposed to PDS measures,results in a slightly degraded precision, but a fargreater completeness in the SPM 3.2. And the useof Hipparcos star proper motions, as opposed toexternal galaxies, places the SPM 3.2 on the ICRSproper-motion system.

6. Kinematics of Metal-poor, Main-sequence

A Stars

In a spectrophotometric survey at the SouthGalactic pole (SGP), Rodgers, Harding and Sadler(1981, hereafter RHS81) identified a group of A-type stars located between 1 and 4 kpc from theGalactic plane and with main-sequence surfacegravities, rather than horizontal-branch ones, asone might expect in an old-halo or thick-disk pop-ulation. Lance (1988a,b, hereafter L88) increasedthe sample, and re-determined radial velocities,calcium abundances, surface gravities, and dis-tances. She confirmed the previous results ofRHS81. The calcium abundance of the main-sequence stars in Lance’s sample ranges between0 and -0.6 dex, and the radial velocity dispersionis 62 km s−1, somewhat higher than the verticalvelocity dispersion of the thick disk (∼ 40 km s−1,e.g., Edvardsson et al. 1993). On these grounds,L88 and RHS81 hypothesized that these stars wereformed in a system accreted by the Milky Way.Rodgers and Roberts (1993) expanded this typeof survey to two additional Galactic locations, inorder to determine the rotational properties of

11

this population from radial velocities alone. Theyfound that the kinematical properties of the A-type, main-sequence stars are similar to those ofthe thick disk. Thus, a new hypothesis was putforward: the stars in question may be thick-diskblue stragglers. However, Rodgers and Roberts(1993) concluded this was unlikely, estimating thatabout a third of the thick disk stars would have tobe blue stragglers, a number that is too large whencompared with the frequency of blue stragglers inglobular clusters.

We have combined the radial-velocity data fromL88 with SPM3 proper motions to determine 3Dvelocities of 29 A-type, main sequence stars at theSGP. There was no overlap between the Rodgersand Roberts (1993) sample and the SPM3 Cat-alog. For comparison, we also examine the A-type, main-sequence stars identified in Wilhelm etal. (1999, hereafter W99), Table 2C, a catalog ofmetal-poor stars. There are 79 stars in commonbetween the SPM3 Catalog and the W99 data set.This latter sample has a metallicity range from-0.1 to -3.0 dex, much wider than that of L88.Thus, the W99 sample is expected to contain asignificant fraction of halo stars.

We have determined velocity components in acylindrical coordinate system, in the Galactic restframe. Π is the radial component (positive out-ward from the Galactic center), Θ is the rota-tional component (positive toward Galactic ro-tation), and W is perpendicular to the Galacticdisk (positive toward the North Galactic Pole). InFig. 9 we show the velocity components of the L88sample, while in Fig. 10 those of W99. The ma-jority of the L88 stars have velocities typical ofthe thick disk, (Θ ≈ 180 km s−1), while the W99velocities are characteristic of both the thick diskand halo. Roughly half of the stars in this lattersample have metallicities lower than -1.0 dex, thetraditional cutoff between thick disk and halo.

The proper-motion results for the L88 sample,thus, support the case for thick-disk blue strag-glers, rather than that for an accreted compo-nent with distinct kinematics. More recently, in aradial-velocity study of blue metal-poor stars, Pre-ston and Sneden (2000) (also supported by Car-ney et al. 2001) found a high fraction of binaries,leading them to suggest that the frequency of bluestragglers in the halo (i.e., field stars) is largerthan that in globular clusters. Their findings, as

do ours, lend support to the notion that distant,A-type, main-sequence stars are more likely Galac-tic field blue stragglers rather than being the rem-nants of a disrupted satellite.

7. Summary

The recently released SPM3 Catalog providespositions, absolute proper motions, and photo-graphic B, V photometry for 10.7 million objects,primarily stars, in an irregular declination bandin the Southern hemisphere. Unlike previous ver-sions of the SPM Catalog, we strive for complete-ness to the limiting magnitude of the plate mate-rial. This is attained by using the Precision Mea-suring Machine of the the US Naval Observatoryto scan the first and second-epoch plates. Theresulting catalog is nearly complete in the range5 < V < 17.5. Also, this is the first SPM Cata-log in which the zero-point of the absolute proper-motion system is defined by Hipparcos stars, asopposed to external galaxies. As with all previousversions of the Catalog, a correction for magnitudeequation is derived and applied to each individ-ual plate. The statistics of the SPM3 Catalog aresummarized in Table 1. The Catalog, along withcross-identifications to other relevant catalogs, isavailable via electronic transfer.

As a brief example of the application of theSPM3 proper motions, we calculate 3D velocitiesfor metal-poor, main-sequence A stars, a groupfirst identified by Rodgers (1971) and subsequentlyexpanded upon. We find these stars’ kinematicsare predominantly similar to that of the thick-disk,lending support for their interpretation as thick-disk blue stragglers, as opposed to being the rem-nants of an accretion process. This is but a single,cursory example from among the large number ofkinematic studies possible with the SPM3 data,which are currently being pursued.

Future versions of the SPM Catalog, expandingcoverage to a larger portion of the Southern sky,will be based on the existing first-epoch photo-graphic plate material and second-epoch CCD ob-servations now underway. A CCD camera systemis currently operating on the double astrographof the Cesco Observatory and providing the nec-essary second-epoch SPM material. The systemconsists of an astrometric/photometric 4K×4KPixelVision camera mounted in the visual focal

12

plane and a photometric 1K×1K Apogee camerain the blue focal plane. Specific regions of interest,e.g. globular-cluster fields and Milky Way satel-lite fields, are expected to be the first CCD-basedSPM projects, with the continuation of the surveyportion of the Southern Proper Motion programto follow.

The SPM3 Catalog would not be possible with-out the efforts of the numerous individuals whomade the first and second-epoch observations atCesco Observatory, and those who performed thethe PMM scanning at USNO-Flagstaff. We aregrateful to them all. We also are grateful for thecollaborative efforts of Tim Beers, who promptedus to construct this version of the Catalog, whileawaiting the accumulation of CCD data neces-sary for future versions. This project has alsobenefited from the assistance and helpful com-ments of Norbert Zacharias. Finally, insightfulcomments from Bob Hanson, who acted as ref-eree to an earlier version of this paper, were ex-tremely helpful in improving not only this paper,but the SPM3 Catalog itself. This publicationmakes use of data from the Two Micron All SkySurvey, (2MASS), which is a joint project of theUniversity of Massachusetts and the Infrared Pro-cessing and Analysis Center/California Instituteof Technology, funded by the National Aeronau-tics and Space Administration and the NationalScience Foundation. The Southern Proper Mo-tion program has been funded over the years by anumber of grants from the National Science Foun-dation, including the most recent, AST-0098548.

REFERENCES

Carney, B. W., Latham, D. W., Laird, J. B., andMorse, J. A. 2001, AJ, 122, 3419

Edvardsson, B., Andersen, B., Gustafson, B.,Lambert, D. L., Nissen, P. E. and Tomkin, J.1993, A&A, 275, 101

The Hipparcos and Tycho Catalogues, ESA SP-1200

Girard, T. M., Platais, I., Kozhurina-Platais, V.,van Altena, W. F., & Lopez, C. E. 1998, AJ,115, 855 (Paper I)

Hog E., Fabricius C., Makarov V. V., Urban S.,Corbin T., Wycoff G., Bastian U., Schwek-endiek P., & Wicenec A. 2000, A&A, 355, L27

Klemola, A. R., Jones, B. F., & Hanson, R. B.1987, AJ, 94, 501

Lance, C. M. 1988a, ApJS, 68, 463

Lance, C. M. 1988b, ApJ, 334, 927

Lasker, B. M., Sturch, C. R., Lopez, C., Malla-mas, A. D., McLaughlin, S. F., Russell, J. L.,Wisniewski, W. Z., Gillespie, B. A., Jenkner,H., Siciliano, E. D., Kenny, D., Baumert, J. H.,Goldberg, A. M., Henry, G. W., Kemper, E., &Siegel, M. J. 1988, ApJS, 68, 1

Monet, D. G., Levine, S. E., Canzian, B., Ables,H. D., Bird, A. R., Dahn, C. C., Guetter, H.H., Harris, H. C., Henden, A. A., Leggett, S.K., & 19 coauthors 2003, AJ, 125, 984

Platais, I., Girard, T. M., Kozhurina-Platais, V.,van Altena, W. F., Lopez, C. E., Mendez, R.A., Ma, W.-Z., Yang, T.-G., MacGillivray, H.T., & Yentis, D. J. 1998, AJ, 116, 2556 (PaperII)

Platais, I., Girard, T. M., van Altena, W. F., Ma,W. Z., Lindegren, L., Crifo, F., & Jahreiß, H.1995, A&A, 304, 141

Preston, G. W. and Sneden, C. 2000, AJ, 120,1014

Rodgers, A. W. 1971, ApJ, 165, 581

Rodgers, A. W., Harding, P., and Sadler, E. 1981,ApJ, 244, 912

Rodgers, A. W., and Roberts, W. H. 1993, AJ,106, 1839

van Altena, W. F., Girard, T., Lopez, C. E., &Molina, E. 1990, in IAU Symp. 141, InertialCoordinate System on the Sky, ed. J. H. Lieske& V. K. Abalakin (Dordrecht: Kluwert), 419

van Altena, W. F., Platais, I., Girard, T. M., &Lopez, C. E. 1994, in Galactic and Solar SystemOptical Astrometry, Proceedings of the RoyalGreenwich Observatory and the Institute of As-tronomy Workshop, held in Cambridge, June21-24, 1993. ed. L. V. Morrison & G. F. Gilmore(Cambridge: Cambridge University Press), 26

Wilhelm, R., Beers, T. C., Sommer-Larsen, J.,Pier, J. R., Layden, A. C., Flynn, C., Rossi,S., and Christensen, P. R. 1999, AJ, 117, 2329

13

Zacharias, N., Urban, S. E., Zacharias, M. I., Hall,D. M., Wycoff, G. L., Rafferty, T. J., Ger-main, M. E., Holdenried, E. R., Pohlman, J.W., Gauss, F. S., Monet, D. G., & Winter, L.2000, AJ, 120, 2131

This 2-column preprint was prepared with the AAS LATEXmacros v5.2.

14

Table 1. SPM3 Statistics.

Sky coverage (sq. deg.) ∼ 3700Magnitude range 4 . V . 17.5Total no. of SPM3 objects 10,764,865Number common to...SPM2 301,2722MASS point sources 10,619,4042MASS extended sources 113,510Tycho-2 166,757Hipparcos 10,901

Fig. 1.— Sky coverage of the SPM, in equatorial coordinates. The gray area indicates the first-epoch platecoverage. The dark area shows the extent of the SPM3 Catalog. Those fields with second-epoch plates, butnot included in the SPM3 Catalog, are indicated by hollow frames.

15

Fig. 2.— Differences in positions, proper motions, and photometry, SPM2 minus preliminary SPM3, beforerecalibration with the SPM2 data. Differences along right ascension and in B magnitudes are shown. Thetrends in the declination differences and in V magnitudes are of similar size and appearance. The samplefield shown is # 333.

16

Fig. 3.— Differences in positions, proper motions, and photometry, SPM2 minus SPM3, after recalibrationwith the SPM2 data. The sample field is the same as in Fig. 2, # 333.

17

Fig. 4.— Proper-motion and position correction masks. a) Proper-motion residuals, Tycho-2 minus SPM3(after recalibration with the SPM2 data), are stacked and averaged as a function of plate coordinates for all156 fields to construct the proper-motion correction mask. b) Position residuals, UCAC 0.9 minus SPM3 (af-ter recalibration with the SPM2 data), are similarly stacked and averaged to construct the position-correctionmask. The plate boundary indicated is 6.4 degrees on a side. Proper-motion and position correction vectorsof 10 mas yr−1 and 100 mas, respectively, are shown for scale.

18

Fig. 5.— Corrections to absolute proper motion. The mean difference between Hipparcos proper motionand preliminary SPM3 proper motions on the system of the SPM2, defines the individual correction for eachfield. The mean correction in (µαcosδ, µδ) for the 156 fields is (0.00, -0.55) mas yr−1 with dispersion (1.12,1.18) mas yr−1.

19

Fig. 6.— Estimated uncertainties in position (top panel), proper motion (middle panel), and V magnitude(lower panel). The points show only a subsample of the Catalog estimated uncertainties, for the sake ofclarity. In the top two panels, the solid curve shows the median for the entire Catalog. Comparisons withthe Hipparcos Catalog motions and the proper motions of identified QSO’s give external error estimates,as shown in the middle panel. The horizontal bars are indicative of the V dispersion of these two samples.The photometry uncertainties are heavily quantized, due to the adoption of a 0.25-magnitude base-leveluncertainty in the photometry linearization step, early in the photometry reduction procedure. Thus, amedian curve is not shown in the lower panel. Instead, the dotted curve shows the dispersion of differencesin V with the CCD-determined calibration photometry.

20

Fig. 7.— Magnitude distribution of SPM3. For comparison, the distributions of the Hipparcos Catalog,Tycho-2 Catalog, and SPM2 Catalog are shown, for the same sky coverage.

Fig. 8.— Comparative magnitude distributions for a sample SPM3 field. Using an approximate relationfor V as a function of 2MASS J and K, the distributions of SPM3 and 2MASS Catalogs can be compareddirectly. The completeness of the SPM3 is approximately the same as that of 2MASS for V < 17.5.

21

Fig. 9.— Galactocentric velocities of metal-poor, main-sequence A stars. The sample is that of Lance 1988a,1988b, (L88).

22

Fig. 10.— Galactocentric velocities of metal-poor, main-sequence A stars. The sample is that of Wilhelm etal. 1999, (W99), and covers a wider range in metallicity, i.e. extending to lower values, than the L88 sampleshown in Fig. 9.

23