New book: Mills, David L. Errata
p. 42, Figure 3.3: 
p. 68 (Section 4.2) The "PLL predicts an adjustment yFLL...while the FLL predicts an adjustment yPLL..." should read "The PLL predicts an adjustment yPLL...while the FLL predicts an adjustment yFLL...".
p. 168, step 3: the gk in the formula should read gy.
p. 257. You define ITAR as International Trade in Arms Regulations. The T stands for Traffic, not Trade.
Contributed by Ben Rothke
p. 42. Last statement in the first paragraph should read "The protocol continues as shown in Figure 3.3".
Contributed by Prof. Fernando G. Tinetti, Universidad Nacional de La Plata, Argentina
P 6, P 12. At one point in the past (before this book) I had read about VFOs in NTP and though that the hardware clock really was a VFO. Its the hardware clock plus a software driver that really provides the VFO. Perhaps others were also confused. [The VFO is implemented by the Unix adjtime() kernel function. DLM]
P 23, third paragraph. The sentences starting "The IANA has assigned" should be prefaced by saying this is for IPv6. [The 224.0.0.1 address is assigned for IPv4; the :101 address is assigned for IPv6. DLM]
P 38. Formula 3.1. This formula troubles me, I suspect some explanation would help, as it is I'd argue it to be confusing or wrong. (...) [The equation is a constant plus the first and second derivatives. The T, R and D are considered constants, but actually vary slowly with time. The offset T is in seconds, the rate R is in seconds per second and aging D (the rate R changes with time) is in seconds per second squared. DLM]
P 42, Figure 3.3. Which is A, which is B? I see "Peer B" on top, but don't know if that means B is on top. [See the label near the bottom "Peer A". DLM]. I don't know which are packets 1, 2, 3. (...). [They should have been laelled t1, t2, ... . DLM]
P 59. Reference #4 is missing. Text refers to Reference 4 on page 36. [#4 should be RFC 781. DLM]
P 67. In the equation with the term y + I+yFLL, the second plus sign should be raised out of the subscript.
P 75, Table 4.1. What happens if |theta|==STEP? [Make that >=. DLM]
P 98, Figure 6.5. The caption should read Malarky time offset.
P 97, line 5. Is the residual offset 50 usec rather than 50 nsec? [From more recent measuremts the residual offset is typically 2-5 usec. DLM]
P 105, Figure 6.16 and nearby text. I can't tell if the offsets are measured in ms or usec. [The y axis label in all figures shows the scale, in this case ms. DLM]
P 109. The x axis is meaningless without a better definition of the scale. Is this freq offset in ppm, in msec, usec, nsec, ?? [Which figure? Figure 6.22 y axis is a probability as labeled; Figure 6.23 is a histogram, so the y axis is the number of hits in the bin. DLM].
P 117, bottom of page. Change unsynchronized to synchronized. [The statement in the book is correct. DLM]
P 135, figure 8.9. The x axis changed between figures 8.8/8.10 and 8.9. The large spike moved from time 8.1h to 8.5h. [The x axis in Figures 8.8 and 8.10 is the hour of the experiment; the x axis in Figure 8.9 is the hour of the particular day the spike occured. DLM]
P 141-142. I can't match the y axis scale on Figure 8.15 to the surrounding text. E.g., the mean in Figure 8.15 isn't -754 ns. [Again, the x axis in figures 8.15 and 8.16 are for the day the event occured. The mean of the offset over the day was indeed as claimed. DLM]
P 221. The use of stratum is different than elsewhere in the book. Can the terminology be disambiguated? [Chapter 13 describes stratum as used in the old AT&T time/frequency distribution network, but still relevant in telephone systems today. That use has nothing to do with the use in NTP. DLM]
p. 194, below Equation 12.5. Equation (1.5) should read Equation (12.5).
p. 195, below Equation (12.9). Equations (1) and (2) should read Equations (12.6) and (12.7).
P 235, Figure 14.3. The mobilize(), clear() and md5() routines should end with return's rather than exits.
P 250. Reference 1. Author's name is Rivest.
P 264, paragraph 2. Update. The leap second happened at the end of 2005.
General comments. The figures could be improved. I can't tell which line is which on figure 4.3, for example. Figure 4.5 is worse. solid/Dotted/dashed lines is one technique. Most figures would benefit with added explanatory text. Figure 6.7 is another example. The upper and lower lines are bounds that should be labeled. The line between are the results for Boulder.
The use of NTP in windows XP, and on other dumb devices should be mentioned. This is a one-time sync I believe, no filtering, no discipline. Likewise ntpdate exists on some systems and seems to be a hybrid one-time sync. DEC, Alpha, and VMS are largely dead. Outside of historical interest why mention them?
Contributed by Steve Sommars, Robert Neal and others.
