T.O. 33B-1-1 5-82 b.    Use the ratio d d 2 1 when one area of the test part is accessible for direct measurement. Where: d₂ = the thickness of an area of the test part as measured by mechanical or optical means d₁ = the  thickness  of  the  same  area  as  indicated  by  the  ultrasonic  instrument  calibrated  on  material similar to the test part Example: Assume  an  area  of  a  test  part  is  measured  with  a  micrometer  and  is  0.167  inch  thick  (d₂  = 0.167 inch).  This same area is measured with ultrasonic instrument and gives a reading of 0.133 inch (d₁ = 0.133 inch). c.     d d 2 1 0 167 0 133 1 25 = = = . . . the correction factor. d.     All  ultrasonic  readings  on  the  test  part  are  now  multiplied  by  1.25  to  obtain  the  actual  thickness.  If another area of the test part gives an ultrasonic reading of 0.200 inch, correct this by multiplying by the correction factor: 0.200 inch x 1.25 = 0.250 inch = the actual test part thickness. 5.5.3.3.1 Flat surfaced reference standards may be used for measurements on convex radii of curvature as small as one inch and concave radii of curvature as small as three inches.  Test parts with radii smaller than one inch convex or three inches concave,  require  reference  standards  with  curved  surfaces  with  radii  equal  to  the  test  part  radii,  ±10%.    In  addition, shoes are required (see paragraphs 5.2.2.9.2 and 5.2.2.10). 5.5.3.3.2 The  surface  finish  of  reference  standards  should  be  63  microinches  or  better  if  maximum  accuracy  is  to  be  obtained. Surface roughness introduces errors as shown Table 5-7. 5.5.3.3.3 The thickness of reference standards shall be measured by mechanical or optical means.  The maximum tolerance for these measurements shall be one-half the tolerance permitted for the ultrasonic thickness measurements on the test part. For example, assume a test part thickness will be measured within ±0.005 inch.  The reference standards must then be measured within ±0.0025 inch.  If the test part measurement tolerance is not specified, the reference standards should be measured within ±0.0005 inch or one-half the instrument tolerance, whichever is greater. Table 5-7.  Measurement Error Introduced by Surface Roughness of Reference Standard or Test Part. Surface Finish (microinches) Measurement Error (inch) 0-63 0.0005 63-125 0.002 125-250 0.005 250-500 0.010 500-20000 0.020