In 1995 it was suggested to establish a homepage of IMO in the World Wide Web (WWW). It was clear, that the web would continue to grow and become an ever more important data source for many people around the world. The Internet was not anymore the communication network of a few scientists or playground for a handful of students and computer freaks, but evolved into a communication and information exchange medium for everybody. The percentage of amateur astronomers using the Internet increased rapidly, and access was not anymore restricted to the developed western countries.
In November 1995 the first homepage of IMO was installed at the web server of the University of Technology Chemnitz (Germany). From the beginning it contained sections for each observing method as well as general information about IMO, it's services and members, and latest news from meteoritics. A mailing list for the fast delivery of announcements and observational results completed the online service. At the IMC 1996 a report of the creation and current status of IMO's homepage was given .
When the author left the University of Chemnitz in early 1997 we had to find a new server for the pages. We found a machine in Belgium, where the files could be installed without any costs and restrictions. What remained was the announcement of the new address all over the world. Unfortunately, there is no kind of general directory in the web where one could declare such things. At the 1996 IMC, however, Jean-Marc Wislez proposed the aquisition of an own domain name for IMO, which turned out to be the best solution.
How does it work? For an annual fee of US$ 50 we ordered the domain imo.net (*). In this domain we
then created symbolic addresses like www.imo.net for our web server, ftp.imo.net for the ftp server,
email@example.com for the mailing list and firstname.lastname@example.org for e-mail contacts with IMO
One advantage is, that such intuitive addresses are easier to keep in mind than long names like the old URL http://www.tu-chemnitz.de/~smo/imo. Furthermore, these addresses are not real but aliases for existing internet addresses. When accessing www.imo.net you don't have to care if you should still contact the old site in Germany or the new one in Belgium. All you need to know is the symbolic name. This makes it possible to move the homepage to another server or change private e-mail addresses of the officials without somebody outside noticing it. Last but not least, we now have our own web server running in Belgium, which allows us to run everything much smoother and improve the services for our virtual guests.
In January '97 we moved IMO's homepage from Germany to Belgium, introduced our own domain, and announced the new address via different channels. The new server started to work on February 1. At that time, also a number of mail aliases were activated:email@example.com - the president of IMO
Later, the mailing list of FIDAC (firstname.lastname@example.org) and the corresponding administrator address (email@example.com) were installed under the new domain, too.
Since the last IMC, the welcome page of IMO has been accessed approximately 15,000 times, which
comes down to an average of roughly 40 visits per day. At the same time, the number of IMO-News
subscribers increased by about 50% to currently 216 readers from more than 42 countries.
One side effect of the new server is, that we now have detailed information about which file is when requested from what site. Each request is stored in a log file and looks like the following example:
michael.vatican.va - - [25/Jun/1997:21:54:31 +0200] "GET / HTTP/1.0" 200 2595
So we know, that on June 25, 1997, at 21:54 UT+2h, somebody was sitting at the machine michael.vatikan.va and visiting the welcome page of IMO (/).
The log file enables us to find out, what information is of most interest to our virtual guests. We can determine, were most of our visitors come from and what for they contact us. This may help us to better evaluate our work and to improve IMO's online service in the future.
In the following, the access statistics of our new web server from the first six months of operation (February 1 - July 31, 1997) will be discussed in detail.
The first question was, where our visitors come from. For that, the distribution of the top-level
domains contacting our web site was computed. As for all the following graphs, only accesses to
*.html files where counted, excluding the /cgi-bin directory and double requests in short succession.
So, the statistics are not biased by logos, counters and images.
In the six months we recorded requests from 80 named and 67 unnamed top-level domains. Figure 1 shows the distribution of the 15 most active domains. The busiest were the commercial (com) and network (net) domain. Unfortunately we don't know much about them, since these include different people from all over the world. They are followed by the US education domain (edu), which is used by universities, high schools and other institutions in the United States.
Figure 1: http://www.imo.net - Top-15 Top-Level-Domains
Looking at the top-15 subdomains (figure 2) we notice, that most accesses (about 7% from all requests) came from the Internet provider America Online. We seem to have one or many interested readers in the atext.com domain, followed by CompuServe. Place four is hold by the German Aerospace Research Establishment DLR (dlr.de). This proves how busy the author was maintaining the homepage of IMO, since most of these requests should come from him. :-)
Figure 2: http://www.imo.net - Top-15 Domains
In total we received requests from almost 4,000 subdomains all around the world.
Let's come back to the home location of our visitors by translating the top-level domains into
country names. As we can see from the corresponding graph (figure 3), more than half of all named
requests originate from one of the international domains (com/org/net), where we don't know the
country. However, it is a good guess that the majority of them come from the US.
Place two falls to the United States (edu/gov/us/mil), followed by Germany, the United Kingdom and Canada. Also here we should not forget, that almost a quarter of all requests were left out as they came from internet hosts without symbolic names.
Figure 3: http://www.imo.net - Top-15 Countries
The country distribution of the IMO-News subscribers (figure 4) looks similar.
Figure 4: IMO-News - Top-15 Countries
From figure 3 one could guess, that visitors from N/S America produced most traffic at IMO's web
site. However, as to be seen in figure 5, the number of requests from European countries was even
higher. Again, the biggest number of named requests came from the international domains which
cannot be located precisely. However, 36 European countries rank second with more than 7,000
accesses, and 13 North and South American states rank third. In addition, we counted requests from
19 Asian countries, 3 domains in Australia and Oceania, and 1 from an African country.
Figure 5 reflects only imperfect the distribution of meteor observers around the world. It seems, that the access to online information of IMO is well established in Europe. The United States produce most traffic at our web site, which is only natural due to the large number of observers, the level of 'computerization' and the mother tongue. However, there are only a few American contacts from outside Canada and the US.
The large number of active Asian amateur astronomers is completely underrepresented in this statistic, which is probably because of the language barrier. It might be worth to think about a translation mechanism to make our information better available in these countries.
Figure 5: http://www.imo.net - Continents
The next analysis dealt with the information that was requested. Figure 6 shows the distribution of requests for the different directories of our server.
Figure 6: http://www.imo.net - Directories
Clearly, the home directory was visited most times. Here we have stored the welcome page and general information like the Who is Who of IMO, membership information, a glossary and a list of links to other meteor-related web sites in the world. About 15% of all visitors requested files in the fireball directory. IMO's meteor shower calendars gained much attention, too, followed by the directory for radio observers.
Figure 7 presents the number of accesses to the top-15 files. About a quarter of all requested the welcome page of IMO. Heavily accessed was also the list of external links and the 1997 Meteor Shower Calendar. Among the next files we find the start page for visual observers, the fireball page and the news section. It is worthwile to note, that the online fireball report form, the meteor glossary and IMO's Who is Who rank among the 15 most wanted files.
Figure 7: http://www.imo.net - Top-15 Files
We can clearly see, that the effords of those who maintain the different sections really pay off. It is only natural that the standard techniques of meteor observation gain more attention than others, but also less common methods like radio or video observation are of great interest to the visitors. The enormous number of request for the shower calendar should encourage the editors to continue their valuable work.
Finally, the time distribution of the requests was analysed. As figure 8 shows, we had a fairly constant rate of about 200 accesses each day. However, especially shortly before and during major meteor showers, the number of requests increases drastically.
Figure 8: http://www.imo.net - Date
On April 18 we recorded five times the usual access rate. During the Perseids in August '97 (not shown in this graph) this increase was even stronger. Within 5 days we counted about 10,000 requests, with a maximum of 3,500 (15 times the average) on August 13 alone! Also the type of information accessed in those days of major meteor shower activity changed sligthly. The actual meteor shower calendar, the page for visual observers and the news sections were requested more often.
Interestingly, we counted most requests in the middle of the week (figure 9), whereas the number decreased at the weekends. The graph looks even smoother if we take out the April 18 data, which overrepresent Friday.
Figure 9: http://www.imo.net - Day of Week
Finally, the distribution of countries is reflected in some sense in the time histogram (figure 10). We have an access maximum in the European afternoon and evening hours, a high plateau of requests in the American evening hours, and a drop inbetween. However, minimum and maximum differ only by a factor of 2.5, so the load of the server is essentially constant.
Figure 10: http://www.imo.net - Time
IMO's website has gained growing attention in the last twelve months. It was accessed from online visitors of more than 70 countries all over the world and provided them with information covering virtually all areas of meteor science. The access statistics should encourage us to continue our work, even if there is not always a direct feedback from our guests. We should regularly update those files that are requested most often, and improve the contents of the others to make them more attractive. It is worth to think about translation mechanisms, that allow those who do not speak English to use at least the most important pages.
I would like to thank Marc Gyssens, who provided the storage place of IMO's homepage at his computer at work. Special thanks to Werner Depoorter, who installed all the software and got the web server and mailing list running on time. He currently acts as the hostmaster and deals with all technical problems, without ever having touched the keyboard of Marc's computer! Thanks also to all who maintained their part of the web pages and helped to obtain the domain name: Rainer Arlt, André Knöfel, Marc de Lignie, Jürgen Rendtel, and Jean-Marc Wislez.
 Molau S. (1997), "IMO goes online"
Proceedings of the International Meteor Conference 1996, p.15