Ten tablets of 17-alpha-methyltestosterone (25 mg) (lot # 44507) were selected as a random sample from a bottle containing 100 tablets. The 17-alpha-methyltestosterone was manufactured by Richlyn laboratories and marketed by United Research laboratories Inc. (Bensalem, Pennsylvania). Ten tablets of meclizine hydrochloride (25 mg) (USP batch # 73174) manufactured by Geneva Generics (Broomfield, Colorado) were selected as a random sample as well. The near-IR spectra of the tablets were obtained with an InfraAlyzer Flex near-IR spectrometer (Bran+Luebbe, Elmsford, New York). A 90 degree conical reflector(5) was used to hold the tablets beneath an integrating sphere for near-IR scanning. Five of the meclizine tablets were scanned with the yellow side up in the holder, and the other five with the white side up because the meclizine tablets were heterogeneous (one side was yellow and the other white). The samples were scanned in a darkened room to reduce the noise from stray light. The samples were also kept in air-tight containers to prevent moisture absorption. The samples were scanned at 19 different near-IR wavelengths between 1445 nm and 2348 nm. The data collected from the spectrometer were transferred to the MicroVax II computer (Digital Equipment Corp., Maynard, Massachusetts) using a program written in Speakeasy IV Epsilon (Speakeasy Computing Corp., Chicago, Illinois). The data analysis was done on an IBM 3090-600J parallel vector supercomputer (IBM, Armonk, New York), also using programs written in Speakeasy.
The same tablets were scanned using the acoustic-resonance spectrometer at 128 different frequencies in the range 10 to 50 kHz (see Fig. 1 - A diagram of a tablet being scanned in the acoustic-resonance spectrometer. The arrows show the directions of signal input and output.). A total of 100 ARS spectra were collected from each sample of tablets (each tablet was scanned 10 times). The spectra were averaged in groups as in ref. 2 to provide ten spectra (one for each tablet in the sample) to reduce spectral noise. The tablets were positioned at the vertex of the quartz tube in Fig. 1 for scanning. The swept-frequency acoustic excitation signal was applied to the input transducer at the top left arrow in Fig.1. The acoustic-resonance signals were received at the output transducer at the top left arrow. The output signals were amplified and digitized with a computer. Acoustic and/or near-IR light could be applied to the bottom of the tablet (in the direction of the bottom arrow) using an additional transmitting transducer or optics positioned there.
