REAL DATA(64,29) 
        REAL TOTAL(29,29), W1(29), W2(29), W3(29)
	REAL MEAN(29), MGP(3,29), BETWEEN(3,3), BETW2(3,3), CPROJ(29,3)
	INTEGER GP(64), NOG(3)
	DIMENSION DUM(29)
	DIMENSION IANAL(4,3)
C
	OPEN(UNIT=21,STATUS='OLD',FILE='luminosity.dat')
C
C	   Get input data.
C
	NG = 3
	N = 64
	M = 29
	DO I = 1, N
	   READ(21,100)(DATA(I,J),J=1,M),gp(i)
           data(i,1) = data(i,1)+1.0  ! Fixup to get +ve values.
  100	   FORMAT(15f6.3,/,14f6.3,i6)
	ENDDO
C
	do j = 1, m
	sum= 0.0
	  do i = 1,n
	     sum=sum+data(i,j)
	  enddo
	  write(6,*) ' total:',j,sum
	enddo
C
	
	IERR = 0
	IPRINT = 3
	CALL MDA(N,M,NG,DATA,GP,IPRINT,NOG,MEAN,MGP,TOTAL,BETWEEN,
     X			BETW2,CPROJ,W1,W2,IERR)
	IF (IERR.NE.0) GOTO 9000
C
C
C ...   ANALYSE THE RESULTS
C
	DO I1 = 1,4
	   DO I2 = 1,3
	      IANAL(I1,I2)=0
	   ENDDO
	ENDDO
	DO I = 1, N
	      ICLASS=GP(I)
	      IF(DATA(I,1).LT.0.0) THEN
		    IANAL(2,ICLASS)=IANAL(2,ICLASS)+1
	      ELSE
	            IANAL(1,ICLASS)=IANAL(1,ICLASS)+1
	      ENDIF
	      IF(DATA(I,2).LT.0.0) THEN
	            IANAL(4,ICLASS)=IANAL(4,ICLASS)+1
	      ELSE
	            IANAL(3,ICLASS)=IANAL(3,ICLASS)+1
	      ENDIF
	ENDDO
	WRITE(6,*) ' '
	WRITE(6,*) ' '
	WRITE(6,*) ' IN THE FOLLOWING, THE 3 COLUMNS ARE THE CLASSES,'
	WRITE(6,*) ' s, g, d  RESPECTIVELY.'
	WRITE(6,*) ' '
	WRITE(6,*) ' THE 4 ROWS ARE THE DIFFERENT QUADRANTS,'
	WRITE(6,*) ' ROWS 1, 2  = POSITIVE, NEGATIVE FIRST AXIS,'
	WRITE(6,*) ' ROWS 3, 4  = POSITIVE, NEGATIVE SECOND AXIS.'
	WRITE(6,*) ' '
	WRITE(6,*) ' LOOK FOR: CLASS x BEING OVERWHELMINGLY IN QUADRANT'
	WRITE(6,*) ' DEFINED BY POSITIVE AXIS a, NEGATIVE AXIS b. ETC.'
	WRITE(6,*) ' '
	DO I1 = 1,4
	   WRITE(6,174) (IANAL(I1,I2),I2=1,3)
  174	   FORMAT(3I4)
	ENDDO
C
C
	GOTO 9900
 9000	WRITE (6,*) ' ABNORMAL END: IERR =', IERR
 9900	CONTINUE
	END