Japanese cedar (Cryptomeria japonica D. Don) was used as the test material. Test specimen dimensions were 100 mm (longitudinal) × 100 mm (radial) × 100 mm (tangential). Numbers of test specimens were 3 pieces (S1, S2, S3). The air-dried density and moisture content show in Table 1.

An air-coupled ultrasonic wave was propagated through air and specimens of wood, as shown in Figure 1. Ultrasonic wave velocities were measured by using a pulser-receiver (JPN-10CKN, Japan probe Co. Ltd., Japan), a preamplifier, and monolithic composite transducers of type 14 × 20 mm with a natural frequency of 200 kHz (Japan probe Co. Ltd., Japan). The propagation directions of the ultrasonic wave corresponded to the axial (longitudinal) and lateral (radial or tangential) directions in wood. Ultrasonic velocity was measured at nine points on the surface of cross and tangential sections, respectively (Figure 2). The room temperature (t) and velocity measurement were recorded at the same time. The ultrasonic velocity (V) was calculated using Equations (1) and (2).

where L is the propagation distance of wood, T is the propagation time with a wood sample, T_a is the propagation time without a wood sample, and V_a is the velocity in air.

In addition, the ultrasonic wave velocities were measured using the contact method. The measurement equip- ments were the same as those used in the non-contact method. The ultrasonic velocity (V_c) was calculated using Equation (3).

where L is the propagation distance of wood, T_c is the propagation time with a wood sample, and T₀ is the propagation time without a wood sample.

Figure 3 shows the receiving waveform in air at 110 mm distance between transducers. The propagation time was determined using a zero-crossing method. The propagation time showed about 330 µs. Figure 4 shows the examples of receiving waveforms in axial and lateral directions of wood. The propagation time in wood is smaller than that in air, because an ultrasonic velocity in wood is faster than that in air. The averaged propagation time in axial and lateral directions was about 60 µs and 100 µs, respectively. The propagation time was ranked in the ascending order of lateral and axial directions. These propagation times were substituted for Equation (1) in order to calculate the velocities. As shown in Figure 3, some small signals that were observed before the receiving waveform are used to determine the propagation time. The small signals may have some important information for ultrasonic propagation characteristics. Further research is required to study these small signals.

Table 2 shows the ultrasonic wave velocities in the axial and lateral directions for three test specimens (S1, S2, S3). The average values of velocity in the axial direction (V_L) were 3964 m/s, 3515 m/s, and 3468 m/s, respectively. Figure 5 shows the values of ultrasonic wave velocity each position in the axial and lateral directions. The minimum values for V_L show 3597 m/s, 3234 m/s, and 2733 m/s, respectively. Their values existed around

^aMeasurement by Hasegawa et al. [2] , ^bMeasurement by Mishiro [14] , ^cVelocity in the radial direction, ^dVelocity in the tangential direction.

the pith. On the other hands, the maximum values show 4414 m/s, 3816 m/s, and 3796 m/s, respectively. Their values existed in the outside point of measurement area. In general, the tracheid length gradually increases toward the outside and attains a constant value. An ultrasonic wave dissipates acoustical energy when it occurs at the end of a fiber [10] . Hasegawa et al. demonstrated that V_L for Japanese cedar and Japanese cypress were significantly related to the tracheid length [2] . Tracheid length seems to affect V_L. On the other hand, the average values of velocity in the lateral direction (V_RT) were 1412 m/s, 1606 m/s, 1331 m/s, respectively. The ratios of V_L to V_RT ranged from 2.2 to 2.8. Distinct relationship among measurement points such as the axial directions was not observed. As shown in Figure 2, the growth ring structure significantly influences the wave propagation. In the center of the test specimens, the propagation path coincides with the radial direction. However, at the edge of

the test specimens, the propagation path coincides with the radial and tangential directions. As a result, the ultrasonic beam is deflected and shifted laterally with respect to the incident angle [11] . In addition, for the measurement of V_RT, the distance between the test specimen and sensor was within the near field distance. As a result, there is a large variation of V_RT in wood. In the future, we will determine the optimal measurement condition such as near-field distance for Japanese cedar column.

As shown in Table 2, the velocities obtained using the contact method (V_c) were larger than those obtained using the non-contact method (V) at a significant level of 5%. The correlation coefficients in the axial and lateral directions were 0.48 and 0.40, respectively. The ratios of V_L to V_RT ranged from 2.4 to 2.5. The ratios of velocity obtained using the contact method were similar to those obtained using the non-contact method. Vun et al. demonstrated that the velocities measured using the non-contact method showed higher values than those measured using the contact method for an oriented strand board [12] . Raffaella et al. reported that the non-contact ultrasonic velocity of food items was significantly higher than the velocity measured using the contact method (P < 0.05) [13] . Results in this study were not in agreement with those in the previous study [12] [13] . Further research is needed to clarify the reason V_c shows larger values than V.

The values of velocity in this study were smaller than those in the previous study, which used a contact method, as shown in Table 2. An ultrasonic wave velocity changes with the specimen dimensions and natural frequency [10] . Bucur demonstrated that the ratios of V_L to V_R in European wood species ranged from 2.4 to 3.4 for softwood [10] . For Japanese cedar, Hasegawa et al. [2] and Mishiro [14] reported that the ratios of velocity in the longitudinal direction to that in the radial direction were 2.3; the ratios of velocity in the longitudinal direction to that in the tangential direction were 3.1. The ratios for a non-contact method were almost the same as those for a contact method. The findings in this study suggest that the ultrasonic wave velocities in Japanese cedar could be measured with a non-contact method by using air-coupled ultrasonics.